★ Charting and Navigating Hugging Face's Model Atlas

Eliahu Horwitz, Nitzan Kurer, Jonathan Kahana, Liel Amar, Yedid Hoshen

As there are now millions of publicly available neural networks, searching and analyzing large model repositories becomes increasingly important. Navigating so many models requires an atlas, but as most models are poorly documented charting such an atlas is challenging. To explore the hidden potential of model repositories, we chart a preliminary atlas representing the documented fraction of Hugging Face. It provides stunning visualizations of the model landscape and evolution. We demonstrate several applications of this atlas including predicting model attributes (e.g., accuracy), and analyzing trends in computer vision models. However, as the current atlas remains incomplete, we propose a method for charting undocumented regions. Specifically, we identify high-confidence structural priors based on dominant real-world model training practices. Leveraging these priors, our approach enables accurate mapping of previously undocumented areas of the atlas. We publicly release our datasets, code, and interactive atlas.

★ SciVerse: Unveiling the Knowledge Comprehension and Visual Reasoning of LMMs on Multi-modal Scientific Problems

Ziyu Guo, Ray Zhang, Hao Chen, Jialin Gao, Dongzhi Jiang, Jiaze Wang, Pheng-Ann Heng

The rapid advancement of Large Multi-modal Models (LMMs) has enabled their application in scientific problem-solving, yet their fine-grained capabilities remain under-explored. In this paper, we introduce SciVerse, a multi-modal scientific evaluation benchmark to thoroughly assess LMMs across 5,735 test instances in five distinct versions. We aim to investigate three key dimensions of LMMs: scientific knowledge comprehension, multi-modal content interpretation, and Chain-of-Thought (CoT) reasoning. To unveil whether LMMs possess sufficient scientific expertise, we first transform each problem into three versions containing different levels of knowledge required for solving, i.e., Knowledge-free, -lite, and -rich. Then, to explore how LMMs interpret multi-modal scientific content, we annotate another two versions, i.e., Vision-rich and -only, marking more question information from texts to diagrams. Comparing the results of different versions, SciVerse systematically examines the professional knowledge stock and visual perception skills of LMMs in scientific domains. In addition, to rigorously assess CoT reasoning, we propose a new scientific CoT evaluation strategy, conducting a step-wise assessment on knowledge and logical errors in model outputs. Our extensive evaluation of different LMMs on SciVerse reveals critical limitations in their scientific proficiency and provides new insights into future developments. Project page: https://sciverse-cuhk.github.io

comment: Initially released in September 2024. Project page: https://sciverse-cuhk.github.io

★ Transformers without Normalization CVPR 2025

Jiachen Zhu, Xinlei Chen, Kaiming He, Yann LeCun, Zhuang Liu

Normalization layers are ubiquitous in modern neural networks and have long been considered essential. This work demonstrates that Transformers without normalization can achieve the same or better performance using a remarkably simple technique. We introduce Dynamic Tanh (DyT), an element-wise operation $DyT($x$) = \tanh(\alpha $x$)$, as a drop-in replacement for normalization layers in Transformers. DyT is inspired by the observation that layer normalization in Transformers often produces tanh-like, $S$-shaped input-output mappings. By incorporating DyT, Transformers without normalization can match or exceed the performance of their normalized counterparts, mostly without hyperparameter tuning. We validate the effectiveness of Transformers with DyT across diverse settings, ranging from recognition to generation, supervised to self-supervised learning, and computer vision to language models. These findings challenge the conventional understanding that normalization layers are indispensable in modern neural networks, and offer new insights into their role in deep networks.

comment: CVPR 2025; Project page: https://jiachenzhu.github.io/DyT/

★ Siege: Autonomous Multi-Turn Jailbreaking of Large Language Models with Tree Search ICLR 2025

Andy Zhou

We introduce Siege, a multi-turn adversarial framework that models the gradual erosion of Large Language Model (LLM) safety through a tree search perspective. Unlike single-turn jailbreaks that rely on one meticulously engineered prompt, Siege expands the conversation at each turn in a breadth-first fashion, branching out multiple adversarial prompts that exploit partial compliance from previous responses. By tracking these incremental policy leaks and re-injecting them into subsequent queries, Siege reveals how minor concessions can accumulate into fully disallowed outputs. Evaluations on the JailbreakBench dataset show that Siege achieves a 100% success rate on GPT-3.5-turbo and 97% on GPT-4 in a single multi-turn run, using fewer queries than baselines such as Crescendo or GOAT. This tree search methodology offers an in-depth view of how model safeguards degrade over successive dialogue turns, underscoring the urgency of robust multi-turn testing procedures for language models.

comment: Accepted to ICLR 2025 Trustworthy LLM

★ From TOWER to SPIRE: Adding the Speech Modality to a Text-Only LLM

Kshitij Ambilduke, Ben Peters, Sonal Sannigrahi, Anil Keshwani, Tsz Kin Lam, Bruno Martins, Marcely Zanon Boito, André F. T. Martins

Large language models (LLMs) have shown remarkable performance and generalization capabilities across multiple languages and tasks, making them very attractive targets for multi-modality integration (e.g., images or speech). In this work, we extend an existing LLM to the speech modality via speech discretization and continued pre-training. In particular, we are interested in multilingual LLMs, such as TOWER, as their pre-training setting allows us to treat discretized speech input as an additional translation language. The resulting open-source model, SPIRE, is able to transcribe and translate English speech input while maintaining TOWER's original performance on translation-related tasks, showcasing that discretized speech input integration as an additional language is feasible during LLM adaptation. We make our code and models available to the community.

★ Compositional Subspace Representation Fine-tuning for Adaptive Large Language Models ICLR 2025

Andy Zhou

Adapting large language models to multiple tasks can cause cross-skill interference, where improvements for one skill degrade another. While methods such as LoRA impose orthogonality constraints at the weight level, they do not fully address interference in hidden-state representations. We propose Compositional Subspace Representation Fine-tuning (CS-ReFT), a novel representation-based approach that learns multiple orthonormal subspace transformations, each specializing in a distinct skill, and composes them via a lightweight router. By isolating these subspace edits in the hidden state, rather than weight matrices, CS-ReFT prevents cross-task conflicts more effectively. On the AlpacaEval benchmark, applying CS-ReFT to Llama-2-7B achieves a 93.94% win rate, surpassing GPT-3.5 Turbo (86.30%) while requiring only 0.0098% of model parameters. These findings show that specialized representation edits, composed via a simple router, significantly enhance multi-task instruction following with minimal overhead.

comment: Accepted to ICLR 2025 SCOPE

★ TruthPrInt: Mitigating LVLM Object Hallucination Via Latent Truthful-Guided Pre-Intervention

Jinhao Duan, Fei Kong, Hao Cheng, James Diffenderfer, Bhavya Kailkhura, Lichao Sun, Xiaofeng Zhu, Xiaoshuang Shi, Kaidi Xu

Object Hallucination (OH) has been acknowledged as one of the major trustworthy challenges in Large Vision-Language Models (LVLMs). Recent advancements in Large Language Models (LLMs) indicate that internal states, such as hidden states, encode the "overall truthfulness" of generated responses. However, it remains under-explored how internal states in LVLMs function and whether they could serve as "per-token" hallucination indicators, which is essential for mitigating OH. In this paper, we first conduct an in-depth exploration of LVLM internal states in relation to OH issues and discover that (1) LVLM internal states are high-specificity per-token indicators of hallucination behaviors. Moreover, (2) different LVLMs encode universal patterns of hallucinations in common latent subspaces, indicating that there exist "generic truthful directions" shared by various LVLMs. Based on these discoveries, we propose Truthful-Guided Pre-Intervention (TruthPrInt) that first learns the truthful direction of LVLM decoding and then applies truthful-guided inference-time intervention during LVLM decoding. We further propose ComnHallu to enhance both cross-LVLM and cross-data hallucination detection transferability by constructing and aligning hallucination latent subspaces. We evaluate TruthPrInt in extensive experimental settings, including in-domain and out-of-domain scenarios, over popular LVLMs and OH benchmarks. Experimental results indicate that TruthPrInt significantly outperforms state-of-the-art methods. Codes will be available at https://github.com/jinhaoduan/TruthPrInt.

comment: 15 pages, 9 figures, the first two authors contributed equally

★ VisualWebInstruct: Scaling up Multimodal Instruction Data through Web Search

Yiming Jia, Jiachen Li, Xiang Yue, Bo Li, Ping Nie, Kai Zou, Wenhu Chen

Vision-Language Models have made significant progress on many perception-focused tasks, however, their progress on reasoning-focused tasks seem to be limited due to the lack of high-quality and diverse training data. In this work, we aim to address the scarcity issue of reasoning-focused multimodal datasets. We propose VisualWebInstruct - a novel approach that leverages search engine to create a diverse, and high-quality dataset spanning multiple disciplines like math, physics, finance, chemistry, etc. Starting with meticulously selected 30,000 seed images, we employ Google Image search to identify websites containing similar images. We collect and process the HTMLs from over 700K unique URL sources. Through a pipeline of content extraction, filtering and synthesis, we build a dataset of approximately 900K question-answer pairs, with 40% being visual QA pairs and the rest as text QA pairs. Models fine-tuned on VisualWebInstruct demonstrate significant performance gains: (1) training from Llava-OV-mid shows 10-20% absolute point gains across benchmarks, (2) training from MAmmoTH-VL shows 5% absoluate gain. Our best model MAmmoTH-VL2 shows state-of-the-art performance within the 10B parameter class on MMMU-Pro-std (40.7%), MathVerse (42.6%), and DynaMath (55.7%). These remarkable results highlight the effectiveness of our dataset in enhancing VLMs' reasoning capabilities for complex multimodal tasks.

comment: Technical Report

★ Language Models, Graph Searching, and Supervision Adulteration: When More Supervision is Less and How to Make More More ICLR 2025

Arvid Frydenlund

This work concerns the path-star task, a minimal example of searching over a graph. The graph, $G$, is star-shaped with $D$ arms radiating from a start node, $s$. A language model (LM) is given $G$, $s$, and a target node $t$, which ends one of the arms and is tasked with generating the arm containing $t$. The minimal nature of this task means only a single choice needs to be made: which of the $D$ arms contains $t$? Decoder-only LMs fail to solve this elementary task above $1/D$ chance due to a learned shortcut that absorbs training supervision. We show how this pathology is caused by excess supervision and we present a series of solutions demonstrating that the task is solvable via decoder-only LMs. We find that the task's minimal nature causes its difficulty, as it prevents task decomposition. Our solutions provide insight into the pathology and its implications for LMs trained via next-token prediction.

comment: A reduced version of this work has been accepted to the Workshop on Spurious Correlation and Shortcut Learning: Foundations and Solutions (SCSL) at ICLR 2025. Full version under review

★ The Impact of Item-Writing Flaws on Difficulty and Discrimination in Item Response Theory

Robin Schmucker, Steven Moore

High-quality test items are essential for educational assessments, particularly within Item Response Theory (IRT). Traditional validation methods rely on resource-intensive pilot testing to estimate item difficulty and discrimination. More recently, Item-Writing Flaw (IWF) rubrics emerged as a domain-general approach for evaluating test items based on textual features. However, their relationship to IRT parameters remains underexplored. To address this gap, we conducted a study involving over 7,000 multiple-choice questions across various STEM subjects (e.g., math and biology). Using an automated approach, we annotated each question with a 19-criteria IWF rubric and studied relationships to data-driven IRT parameters. Our analysis revealed statistically significant links between the number of IWFs and IRT difficulty and discrimination parameters, particularly in life and physical science domains. We further observed how specific IWF criteria can impact item quality more and less severely (e.g., negative wording vs. implausible distractors). Overall, while IWFs are useful for predicting IRT parameters--particularly for screening low-difficulty MCQs--they cannot replace traditional data-driven validation methods. Our findings highlight the need for further research on domain-general evaluation rubrics and algorithms that understand domain-specific content for robust item validation.

★ Probing LLMs for Multilingual Discourse Generalization Through a Unified Label Set

Florian Eichin, Yang Janet Liu, Barbara Plank, Michael A. Hedderich

Discourse understanding is essential for many NLP tasks, yet most existing work remains constrained by framework-dependent discourse representations. This work investigates whether large language models (LLMs) capture discourse knowledge that generalizes across languages and frameworks. We address this question along two dimensions: (1) developing a unified discourse relation label set to facilitate cross-lingual and cross-framework discourse analysis, and (2) probing LLMs to assess whether they encode generalizable discourse abstractions. Using multilingual discourse relation classification as a testbed, we examine a comprehensive set of 23 LLMs of varying sizes and multilingual capabilities. Our results show that LLMs, especially those with multilingual training corpora, can generalize discourse information across languages and frameworks. Further layer-wise analyses reveal that language generalization at the discourse level is most salient in the intermediate layers. Lastly, our error analysis provides an account of challenging relation classes.

comment: 18 pages, 7 figures, 3 tables, code: https://github.com/mainlp/discourse_probes

★ MMLU-ProX: A Multilingual Benchmark for Advanced Large Language Model Evaluation

Weihao Xuan, Rui Yang, Heli Qi, Qingcheng Zeng, Yunze Xiao, Yun Xing, Junjue Wang, Huitao Li, Xin Li, Kunyu Yu, Nan Liu, Qingyu Chen, Douglas Teodoro, Edison Marrese-Taylor, Shijian Lu, Yusuke Iwasawa, Yutaka Matsuo, Irene Li

Traditional benchmarks struggle to evaluate increasingly sophisticated language models in multilingual and culturally diverse contexts. To address this gap, we introduce MMLU-ProX, a comprehensive multilingual benchmark covering 13 typologically diverse languages with approximately 11,829 questions per language. Building on the challenging reasoning-focused design of MMLU-Pro, our framework employs a semi-automatic translation process: translations generated by state-of-the-art large language models (LLMs) are rigorously evaluated by expert annotators to ensure conceptual accuracy, terminological consistency, and cultural relevance. We comprehensively evaluate 25 state-of-the-art LLMs using 5-shot chain-of-thought (CoT) and zero-shot prompting strategies, analyzing their performance across linguistic and cultural boundaries. Our experiments reveal consistent performance degradation from high-resource languages to lower-resource ones, with the best models achieving over 70% accuracy on English but dropping to around 40% for languages like Swahili, highlighting persistent gaps in multilingual capabilities despite recent advances. MMLU-ProX is an ongoing project; we are expanding our benchmark by incorporating additional languages and evaluating more language models to provide a more comprehensive assessment of multilingual capabilities.

★ Source-primed Multi-turn Conversation Helps Large Language Models Translate Documents

Hanxu Hu, Jannis Vamvas, Rico Sennrich

LLMs have paved the way for truly simple document-level machine translation, but challenges such as omission errors remain. In this paper, we study a simple method for handling document-level machine translation, by leveraging previous contexts in a multi-turn conversational manner. Specifically, by decomposing documents into segments and iteratively translating them while maintaining previous turns, this method ensures coherent translations without additional training, and can fully re-use the KV cache of previous turns thus minimizing computational overhead. We further propose a `source-primed' method that first provides the whole source document before multi-turn translation. We empirically show this multi-turn method outperforms both translating entire documents in a single turn and translating each segment independently according to multiple automatic metrics in representative LLMs, establishing a strong baseline for document-level translation using LLMs.

comment: 9 pages, 2 figures

★ LLMs in Disease Diagnosis: A Comparative Study of DeepSeek-R1 and O3 Mini Across Chronic Health Conditions

Gaurav Kumar Gupta, Pranal Pande

Large Language Models (LLMs) are revolutionizing medical diagnostics by enhancing both disease classification and clinical decision-making. In this study, we evaluate the performance of two LLM- based diagnostic tools, DeepSeek R1 and O3 Mini, using a structured dataset of symptoms and diagnoses. We assessed their predictive accuracy at both the disease and category levels, as well as the reliability of their confidence scores. DeepSeek R1 achieved a disease-level accuracy of 76% and an overall accuracy of 82%, outperforming O3 Mini, which attained 72% and 75% respectively. Notably, DeepSeek R1 demonstrated exceptional performance in Mental Health, Neurological Disorders, and Oncology, where it reached 100% accuracy, while O3 Mini excelled in Autoimmune Disease classification with 100% accuracy. Both models, however, struggled with Respiratory Disease classification, recording accuracies of only 40% for DeepSeek R1 and 20% for O3 Mini. Additionally, the analysis of confidence scores revealed that DeepSeek R1 provided high-confidence predictions in 92% of cases, compared to 68% for O3 Mini. Ethical considerations regarding bias, model interpretability, and data privacy are also discussed to ensure the responsible integration of LLMs into clinical practice. Overall, our findings offer valuable insights into the strengths and limitations of LLM-based diagnostic systems and provide a roadmap for future enhancements in AI-driven healthcare.

comment: 12 pages, 3 figures

★ World Modeling Makes a Better Planner: Dual Preference Optimization for Embodied Task Planning

Siyin Wang, Zhaoye Fei, Qinyuan Cheng, Shiduo Zhang, Panpan Cai, Jinlan Fu, Xipeng Qiu

Recent advances in large vision-language models (LVLMs) have shown promise for embodied task planning, yet they struggle with fundamental challenges like dependency constraints and efficiency. Existing approaches either solely optimize action selection or leverage world models during inference, overlooking the benefits of learning to model the world as a way to enhance planning capabilities. We propose Dual Preference Optimization (D$^2$PO), a new learning framework that jointly optimizes state prediction and action selection through preference learning, enabling LVLMs to understand environment dynamics for better planning. To automatically collect trajectories and stepwise preference data without human annotation, we introduce a tree search mechanism for extensive exploration via trial-and-error. Extensive experiments on VoTa-Bench demonstrate that our D$^2$PO-based method significantly outperforms existing methods and GPT-4o when applied to Qwen2-VL (7B), LLaVA-1.6 (7B), and LLaMA-3.2 (11B), achieving superior task success rates with more efficient execution paths.

★ Statistical Analysis of Sentence Structures through ASCII, Lexical Alignment and PCA

Abhijeet Sahdev

While utilizing syntactic tools such as parts-of-speech (POS) tagging has helped us understand sentence structures and their distribution across diverse corpora, it is quite complex and poses a challenge in natural language processing (NLP). This study focuses on understanding sentence structure balance - usages of nouns, verbs, determiners, etc - harmoniously without relying on such tools. It proposes a novel statistical method that uses American Standard Code for Information Interchange (ASCII) codes to represent text of 11 text corpora from various sources and their lexical category alignment after using their compressed versions through PCA, and analyzes the results through histograms and normality tests such as Shapiro-Wilk and Anderson-Darling Tests. By focusing on ASCII codes, this approach simplifies text processing, although not replacing any syntactic tools but complementing them by offering it as a resource-efficient tool for assessing text balance. The story generated by Grok shows near normality indicating balanced sentence structures in LLM outputs, whereas 4 out of the remaining 10 pass the normality tests. Further research could explore potential applications in text quality evaluation and style analysis with syntactic integration for more broader tasks.

★ Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyond

Liang Wen, Yunke Cai, Fenrui Xiao, Xin He, Qi An, Zhenyu Duan, Yimin Du, Junchen Liu, Lifu Tang, Xiaowei Lv, Haosheng Zou, Yongchao Deng, Shousheng Jia, Xiangzheng Zhang

This paper presents our work on the Light-R1 series, with models, data, and code all released. We first focus on training long COT models from scratch, specifically starting from models initially lacking long COT capabilities. Using a curriculum training recipe consisting of two-stage SFT and semi-on-policy DPO, we train our model Light-R1-32B from Qwen2.5-32B-Instruct, resulting in superior math performance compared to DeepSeek-R1-Distill-Qwen-32B. Despite being trained exclusively on math data, Light-R1-32B shows strong generalization across other domains. In the subsequent phase of this work, we highlight the significant benefit of the 3k dataset constructed for the second SFT stage on enhancing other models. By fine-tuning DeepSeek-R1-Distilled models using this dataset, we obtain new SOTA models in 7B and 14B, while the 32B model, Light-R1-32B-DS performed comparably to QwQ-32B and DeepSeek-R1. Furthermore, we extend our work by applying reinforcement learning, specifically GRPO, on long-COT models to further improve reasoning performance. We successfully train our final Light-R1-14B-DS with RL, achieving SOTA performance among 14B parameter models in math. With AIME24 & 25 scores of 74.0 and 60.2 respectively, Light-R1-14B-DS surpasses even many 32B models and DeepSeek-R1-Distill-Llama-70B. Its RL training also exhibits well expected behavior, showing simultaneous increase in response length and reward score. The Light-R1 series of work validates training long-COT models from scratch, showcases the art in SFT data and releases SOTA models from RL.

comment: all release at https://github.com/Qihoo360/Light-R1

★ DynaCode: A Dynamic Complexity-Aware Code Benchmark for Evaluating Large Language Models in Code Generation

Wenhao Hu, Jinhao Duan, Chunchen Wei, Li Zhang, Yue Zhang, Kaidi Xu

The rapid advancement of large language models (LLMs) has significantly improved their performance in code generation tasks. However, existing code benchmarks remain static, consisting of fixed datasets with predefined problems. This makes them vulnerable to memorization during training, where LLMs recall specific test cases instead of generalizing to new problems, leading to data contamination and unreliable evaluation results. To address these issues, we introduce DynaCode, a dynamic, complexity-aware benchmark that overcomes the limitations of static datasets. DynaCode evaluates LLMs systematically using a complexity-aware metric, incorporating both code complexity and call-graph structures. DynaCode achieves large-scale diversity, generating up to 189 million unique nested code problems across four distinct levels of code complexity, referred to as units, and 16 types of call graphs. Results on 12 latest LLMs show an average performance drop of 16.8% to 45.7% compared to MBPP+, a static code generation benchmark, with performance progressively decreasing as complexity increases. This demonstrates DynaCode's ability to effectively differentiate LLMs. Additionally, by leveraging call graphs, we gain insights into LLM behavior, particularly their preference for handling subfunction interactions within nested code.

comment: 16 pages, 11 figures

★ BeamLLM: Vision-Empowered mmWave Beam Prediction with Large Language Models

Can Zheng, Jiguang He, Guofa Cai, Zitong Yu, Chung G. Kang

In this paper, we propose BeamLLM, a vision-aided millimeter-wave (mmWave) beam prediction framework leveraging large language models (LLMs) to address the challenges of high training overhead and latency in mmWave communication systems. By combining computer vision (CV) with LLMs' cross-modal reasoning capabilities, the framework extracts user equipment (UE) positional features from RGB images and aligns visual-temporal features with LLMs' semantic space through reprogramming techniques. Evaluated on a realistic vehicle-to-infrastructure (V2I) scenario, the proposed method achieves 61.01% top-1 accuracy and 97.39% top-3 accuracy in standard prediction tasks, significantly outperforming traditional deep learning models. In few-shot prediction scenarios, the performance degradation is limited to 12.56% (top-1) and 5.55% (top-3) from time sample 1 to 10, demonstrating superior prediction capability.

comment: 6 pages, 7 figures, conference

★ VisTai: Benchmarking Vision-Language Models for Traditional Chinese in Taiwan

Zhi Rui Tam, Ya-Ting Pai, Yen-Wei Lee

In this paper, we propose a comprehensive evaluation benchmark for Visual Language Models (VLM) in Traditional Chinese. Our evaluation suite, the first of its kind, contains two complementary components: (1) VisTai-MCQ, a collection of manually curated exam multi-choice questions from 21 academic subjects designed to test the broad knowledge and reasoning capabilities of VLMs; and (2) VisTai-Dialogue, an open dialogue benchmark comprising 131 image-question pairs manually created to evaluate VLMs' ability in free-form dialogue generation within Taiwanese cultural contexts. These benchmarks address a critical gap in the evaluation landscape, where existing benchmarks predominantly focus on English or Simplified Chinese, neglecting the unique linguistic and cultural aspects of Traditional Chinese used in regions like Taiwan and Hong Kong. Our analysis reveals significant performance differences across various VLMs and highlights specific challenges in processing Traditional Chinese visual content.

★ Understanding the Logical Capabilities of Large Language Models via Out-of-Context Representation Learning

Jonathan Shaki, Emanuele La Malfa, Michael Wooldridge, Sarit Kraus

We study the capabilities of Large Language Models (LLM) on binary relations, a ubiquitous concept in math employed in most reasoning, math and logic benchmarks. This work focuses on equality, inequality, and inclusion, along with the properties they satisfy, such as ir/reflexivity, a/symmetry, transitivity, and logical complexity (e.g., number of reasoning ``hops''). We propose an alternative to in-context learning that trains only the representations of newly introduced tokens, namely out-of-context representation learning. This method mitigates linguistic biases already present in a model and, differently from in-context learning, does not rely on external information or illustrations. We argue out-of-context representation learning as a better alternative to in-context learning and fine-tuning to evaluate the capabilities of LLMs on logic tasks that are the building blocks of more complex reasoning benchmarks.

★ G-Boost: Boosting Private SLMs with General LLMs

Yijiang Fan, Yuren Mao, Longbin Lai, Ying Zhang, Zhengping Qian, Yunjun Gao

Due to the limited computational resources, most Large Language Models (LLMs) developers can only fine-tune Small Language Models (SLMs) on their own data. These private SLMs typically have limited effectiveness. To boost the performance of private SLMs, this paper proposes to ask general LLMs for help. The general LLMs can be APIs or larger LLMs whose inference cost the developers can afford. Specifically, we propose the G-Boost framework where a private SLM adaptively performs collaborative inference with a general LLM under the guide of process reward. Experiments demonstrate that our framework can significantly boost the performance of private SLMs.

★ Do I look like a `cat.n.01` to you? A Taxonomy Image Generation Benchmark

Viktor Moskvoretskii, Alina Lobanova, Ekaterina Neminova, Chris Biemann, Alexander Panchenko, Irina Nikishina

This paper explores the feasibility of using text-to-image models in a zero-shot setup to generate images for taxonomy concepts. While text-based methods for taxonomy enrichment are well-established, the potential of the visual dimension remains unexplored. To address this, we propose a comprehensive benchmark for Taxonomy Image Generation that assesses models' abilities to understand taxonomy concepts and generate relevant, high-quality images. The benchmark includes common-sense and randomly sampled WordNet concepts, alongside the LLM generated predictions. The 12 models are evaluated using 9 novel taxonomy-related text-to-image metrics and human feedback. Moreover, we pioneer the use of pairwise evaluation with GPT-4 feedback for image generation. Experimental results show that the ranking of models differs significantly from standard T2I tasks. Playground-v2 and FLUX consistently outperform across metrics and subsets and the retrieval-based approach performs poorly. These findings highlight the potential for automating the curation of structured data resources.

comment: Labeled data and generated image Wordnet are published at https://huggingface.co/collections/VityaVitalich/generated-image-wordnet-67d2c868ff1414ec2f8e0d3d

★ A Hybrid Architecture with Efficient Fine Tuning for Abstractive Patent Document Summarization

Nevidu Jayatilleke, Ruvan Weerasinghe

Automatic patent summarization approaches that help in the patent analysis and comprehension procedure are in high demand due to the colossal growth of innovations. The development of natural language processing (NLP), text mining, and deep learning has notably amplified the efficacy of text summarization models for abundant types of documents. Summarizing patent text remains a pertinent challenge due to the labyrinthine writing style of these documents, which includes technical and legal intricacies. Additionally, these patent document contents are considerably lengthier than archetypal documents, which intricates the process of extracting pertinent information for summarization. Embodying extractive and abstractive text summarization methodologies into a hybrid framework, this study proposes a system for efficiently creating abstractive summaries of patent records. The procedure involves leveraging the LexRank graph-based algorithm to retrieve the important sentences from input parent texts, then utilizing a Bidirectional Auto-Regressive Transformer (BART) model that has been fine-tuned using Low-Ranking Adaptation (LoRA) for producing text summaries. This is accompanied by methodical testing and evaluation strategies. Furthermore, the author employed certain meta-learning techniques to achieve Domain Generalization (DG) of the abstractive component across multiple patent fields.

comment: Accepted Paper in the 8th International Research Conference on Smart Computing and Systems Engineering, University of Kelaniya, Sri Lanka. (Pending Publication)

★ New Trends for Modern Machine Translation with Large Reasoning Models

Sinuo Liu, Chenyang Lyu, Minghao Wu, Longyue Wang, Weihua Luo, Kaifu Zhang

Recent advances in Large Reasoning Models (LRMs), particularly those leveraging Chain-of-Thought reasoning (CoT), have opened brand new possibility for Machine Translation (MT). This position paper argues that LRMs substantially transformed traditional neural MT as well as LLMs-based MT paradigms by reframing translation as a dynamic reasoning task that requires contextual, cultural, and linguistic understanding and reasoning. We identify three foundational shifts: 1) contextual coherence, where LRMs resolve ambiguities and preserve discourse structure through explicit reasoning over cross-sentence and complex context or even lack of context; 2) cultural intentionality, enabling models to adapt outputs by inferring speaker intent, audience expectations, and socio-linguistic norms; 3) self-reflection, LRMs can perform self-reflection during the inference time to correct the potential errors in translation especially extremely noisy cases, showing better robustness compared to simply mapping X->Y translation. We explore various scenarios in translation including stylized translation, document-level translation and multimodal translation by showcasing empirical examples that demonstrate the superiority of LRMs in translation. We also identify several interesting phenomenons for LRMs for MT including auto-pivot translation as well as the critical challenges such as over-localisation in translation and inference efficiency. In conclusion, we think that LRMs redefine translation systems not merely as text converters but as multilingual cognitive agents capable of reasoning about meaning beyond the text. This paradigm shift reminds us to think of problems in translation beyond traditional translation scenarios in a much broader context with LRMs - what we can achieve on top of it.

★ KV-Distill: Nearly Lossless Learnable Context Compression for LLMs

Vivek Chari, Guanghui Qin, Benjamin Van Durme

Sequence-to-sequence tasks often benefit from long contexts, but the quadratic complexity of self-attention in standard Transformers renders this non-trivial. During generation, temporary representations -stored in the so-called KV cache-account for a large portion of GPU memory usage and scale linearly with context length. We introduce KV-Distill, a Transformer compression framework that distills long context KV caches into significantly shorter representations in a question-independent fashion. KV-Distill can be trained as a parameter-efficient adaptor for pretrained models, and enables the compression of arbitrary spans of a context while preserving pre-trained model capabilities. We treat a compressed-uncompressed cache as a student-teacher pairing and apply a KL-type divergence to match the generated outputs. KV-Distill outperforms other compression techniques in worst-case extractive tasks and approaches uncompressed performance in long context question answering and summarization, and it can be fine-tuned on domain-specific contexts to reduce lengths by up to 99% while preserving downstream performance. We demonstrate the generalizability of KV-Distill across various model sizes and architectures.

★ OSMa-Bench: Evaluating Open Semantic Mapping Under Varying Lighting Conditions

Maxim Popov, Regina Kurkova, Mikhail Iumanov, Jaafar Mahmoud, Sergey Kolyubin

Open Semantic Mapping (OSM) is a key technology in robotic perception, combining semantic segmentation and SLAM techniques. This paper introduces a dynamically configurable and highly automated LLM/LVLM-powered pipeline for evaluating OSM solutions called OSMa-Bench (Open Semantic Mapping Benchmark). The study focuses on evaluating state-of-the-art semantic mapping algorithms under varying indoor lighting conditions, a critical challenge in indoor environments. We introduce a novel dataset with simulated RGB-D sequences and ground truth 3D reconstructions, facilitating the rigorous analysis of mapping performance across different lighting conditions. Through experiments on leading models such as ConceptGraphs, BBQ and OpenScene, we evaluate the semantic fidelity of object recognition and segmentation. Additionally, we introduce a Scene Graph evaluation method to analyze the ability of models to interpret semantic structure. The results provide insights into the robustness of these models, forming future research directions for developing resilient and adaptable robotic systems. Our code is available at https://be2rlab.github.io/OSMa-Bench/.

comment: Project page: https://be2rlab.github.io/OSMa-Bench/

★ Wikipedia is Not a Dictionary, Delete! Text Classification as a Proxy for Analysing Wiki Deletion Discussions

Hsuvas Borkakoty, Luis Espinosa-Anke

Automated content moderation for collaborative knowledge hubs like Wikipedia or Wikidata is an important yet challenging task due to multiple factors. In this paper, we construct a database of discussions happening around articles marked for deletion in several Wikis and in three languages, which we then use to evaluate a range of LMs on different tasks (from predicting the outcome of the discussion to identifying the implicit policy an individual comment might be pointing to). Our results reveal, among others, that discussions leading to deletion are easier to predict, and that, surprisingly, self-produced tags (keep, delete or redirect) don't always help guiding the classifiers, presumably because of users' hesitation or deliberation within comments.

comment: Accepted to WNUT-2025

★ VisualPRM: An Effective Process Reward Model for Multimodal Reasoning

Weiyun Wang, Zhangwei Gao, Lianjie Chen, Zhe Chen, Jinguo Zhu, Xiangyu Zhao, Yangzhou Liu, Yue Cao, Shenglong Ye, Xizhou Zhu, Lewei Lu, Haodong Duan, Yu Qiao, Jifeng Dai, Wenhai Wang

We introduce VisualPRM, an advanced multimodal Process Reward Model (PRM) with 8B parameters, which improves the reasoning abilities of existing Multimodal Large Language Models (MLLMs) across different model scales and families with Best-of-N (BoN) evaluation strategies. Specifically, our model improves the reasoning performance of three types of MLLMs and four different model scales. Even when applied to the highly capable InternVL2.5-78B, it achieves a 5.9-point improvement across seven multimodal reasoning benchmarks. Experimental results show that our model exhibits superior performance compared to Outcome Reward Models and Self-Consistency during BoN evaluation. To facilitate the training of multimodal PRMs, we construct a multimodal process supervision dataset VisualPRM400K using an automated data pipeline. For the evaluation of multimodal PRMs, we propose VisualProcessBench, a benchmark with human-annotated step-wise correctness labels, to measure the abilities of PRMs to detect erroneous steps in multimodal reasoning tasks. We hope that our work can inspire more future research and contribute to the development of MLLMs. Our model, data, and benchmark are released in https://internvl.github.io/blog/2025-03-13-VisualPRM/.

★ An Expanded Massive Multilingual Dataset for High-Performance Language Technologies

Laurie Burchell, Ona de Gibert, Nikolay Arefyev, Mikko Aulamo, Marta Bañón, and Pinzhen Chen, Mariia Fedorova, Liane Guillou, Barry Haddow, Jan Hajič, and Jindřich Helcl, Erik Henriksson, Mateusz Klimaszewski, Ville Komulainen, and Andrey Kutuzov, Joona Kytöniemi, Veronika Laippala, Petter Mæhlum, and Bhavitvya Malik, Farrokh Mehryary, Vladislav Mikhailov, Nikita Moghe, and Amanda Myntti, Dayyán O'Brien, Stephan Oepen, Proyag Pal, Jousia Piha, and Sampo Pyysalo, Gema Ramírez-Sánchez, David Samuel, Pavel Stepachev, and Jörg Tiedemann, Dušan Variš, Tereza Vojtěchová, Jaume Zaragoza-Bernabeu

Training state-of-the-art large language models requires vast amounts of clean and diverse textual data. However, building suitable multilingual datasets remains a challenge. In this work, we present HPLT v2, a collection of high-quality multilingual monolingual and parallel corpora. The monolingual portion of the data contains 8T tokens covering 193 languages, while the parallel data contains 380M sentence pairs covering 51 languages. We document the entire data pipeline and release the code to reproduce it. We provide extensive analysis of the quality and characteristics of our data. Finally, we evaluate the performance of language models and machine translation systems trained on HPLT v2, demonstrating its value.

★ MinorBench: A hand-built benchmark for content-based risks for children

Shaun Khoo, Gabriel Chua, Rachel Shong

Large Language Models (LLMs) are rapidly entering children's lives - through parent-driven adoption, schools, and peer networks - yet current AI ethics and safety research do not adequately address content-related risks specific to minors. In this paper, we highlight these gaps with a real-world case study of an LLM-based chatbot deployed in a middle school setting, revealing how students used and sometimes misused the system. Building on these findings, we propose a new taxonomy of content-based risks for minors and introduce MinorBench, an open-source benchmark designed to evaluate LLMs on their ability to refuse unsafe or inappropriate queries from children. We evaluate six prominent LLMs under different system prompts, demonstrating substantial variability in their child-safety compliance. Our results inform practical steps for more robust, child-focused safety mechanisms and underscore the urgency of tailoring AI systems to safeguard young users.

★ ARLED: Leveraging LED-based ARMAN Model for Abstractive Summarization of Persian Long Documents

Samira Zangooei, Amirhossein Darmani, Hossein Farahmand Nezhad, Laya Mahmoudi

The increasing volume of textual data poses challenges in reading and comprehending large documents, particularly for scholars who need to extract useful information from research articles. Automatic text summarization has emerged as a powerful tool to condense lengthy documents into concise and informative summaries. Depending on the approach used, text summarization can be categorized as either extractive or abstractive. While extractive methods are commonly used due to their simplicity, they often miss important information. On the other hand, Abstractive Summarization can generate more coherent and informative summaries by understanding the underlying meaning of the text. Abstractive techniques have gained attention in various languages, and recent advancements have been achieved through pre-training models such as BERT, BART, and T5. However, the challenge of summarizing long documents remains, and alternative models like Longformer have been introduced to address this limitation. In this context, this paper focuses on abstractive summarization in the Persian language. The authors introduce a new dataset of 300,000 full-text Persian papers obtained from the Ensani website and apply the ARMAN model, based on the Longformer architecture, to generate summaries. The experimental results demonstrate promising performance in Persian text summarization. The paper provides a comprehensive overview of related work, discusses the methodology, presents the experimental results, and concludes with future research directions.

comment: 11 pages, 3 tables

★ R.U.Psycho? Robust Unified Psychometric Testing of Language Models

Julian Schelb, Orr Borin, David Garcia, Andreas Spitz

Generative language models are increasingly being subjected to psychometric questionnaires intended for human testing, in efforts to establish their traits, as benchmarks for alignment, or to simulate participants in social science experiments. While this growing body of work sheds light on the likeness of model responses to those of humans, concerns are warranted regarding the rigour and reproducibility with which these experiments may be conducted. Instabilities in model outputs, sensitivity to prompt design, parameter settings, and a large number of available model versions increase documentation requirements. Consequently, generalization of findings is often complex and reproducibility is far from guaranteed. In this paper, we present R.U.Psycho, a framework for designing and running robust and reproducible psychometric experiments on generative language models that requires limited coding expertise. We demonstrate the capability of our framework on a variety of psychometric questionnaires, which lend support to prior findings in the literature. R.U.Psycho is available as a Python package at https://github.com/julianschelb/rupsycho.

★ Assessing the validity of new paradigmatic complexity measures as criterial features for proficiency in L2 writings in English

Cyriel Mallart, Andrew Simpkin, Nicolas Ballier, Paula Lissón, Rémi Venant, Jen-Yu Li, Bernardo Stearns, Thomas Gaillat

This article addresses Second Language (L2) writing development through an investigation of new grammatical and structural complexity metrics. We explore the paradigmatic production in learner English by linking language functions to specific grammatical paradigms. Using the EFCAMDAT as a gold standard and a corpus of French learners as an external test set, we employ a supervised learning framework to operationalise and evaluate seven microsystems. We show that learner levels are associated with the seven microsystems (MS). Using ordinal regression modelling for evaluation, the results show that all MS are significant but yield a low impact if taken individually. However, their influence is shown to be impactful if taken as a group. These microsystems and their measurement method suggest that it is possible to use them as part of broader-purpose CALL systems focused on proficiency assessment.

★ Adaptive Inner Speech-Text Alignment for LLM-based Speech Translation

Henglyu Liu, Andong Chen, Kehai Chen, Xuefeng Bai, Meizhi Zhong, Yuan Qiu, Min Zhang

Recent advancement of large language models (LLMs) has led to significant breakthroughs across various tasks, laying the foundation for the development of LLM-based speech translation systems. Existing methods primarily focus on aligning inputs and outputs across modalities while overlooking deeper semantic alignment within model representations. To address this limitation, we propose an Adaptive Inner Speech-Text Alignment (AI-STA) method to bridge the modality gap by explicitly aligning speech and text representations at selected layers within LLMs. To achieve this, we leverage the optimal transport (OT) theory to quantify fine-grained representation discrepancies between speech and text. Furthermore, we utilize the cross-modal retrieval technique to identify the layers that are best suited for alignment and perform joint training on these layers. Experimental results on speech translation (ST) tasks demonstrate that AI-STA significantly improves the translation performance of large speech-text models (LSMs), outperforming previous state-of-the-art approaches. Our findings highlight the importance of inner-layer speech-text alignment in LLMs and provide new insights into enhancing cross-modal learning.

comment: 12 pages, 7 figures

★ Red Teaming Contemporary AI Models: Insights from Spanish and Basque Perspectives

Miguel Romero-Arjona, Pablo Valle, Juan C. Alonso, Ana B. Sánchez, Miriam Ugarte, Antonia Cazalilla, Vicente Cambrón, José A. Parejo, Aitor Arrieta, Sergio Segura

The battle for AI leadership is on, with OpenAI in the United States and DeepSeek in China as key contenders. In response to these global trends, the Spanish government has proposed ALIA, a public and transparent AI infrastructure incorporating small language models designed to support Spanish and co-official languages such as Basque. This paper presents the results of Red Teaming sessions, where ten participants applied their expertise and creativity to manually test three of the latest models from these initiatives$\unicode{x2013}$OpenAI o3-mini, DeepSeek R1, and ALIA Salamandra$\unicode{x2013}$focusing on biases and safety concerns. The results, based on 670 conversations, revealed vulnerabilities in all the models under test, with biased or unsafe responses ranging from 29.5% in o3-mini to 50.6% in Salamandra. These findings underscore the persistent challenges in developing reliable and trustworthy AI systems, particularly those intended to support Spanish and Basque languages.

★ PRISM: Preference Refinement via Implicit Scene Modeling for 3D Vision-Language Preference-Based Reinforcement Learning

Yirong Sun, Yanjun Chen

We propose PRISM, a novel framework designed to overcome the limitations of 2D-based Preference-Based Reinforcement Learning (PBRL) by unifying 3D point cloud modeling and future-aware preference refinement. At its core, PRISM adopts a 3D Point Cloud-Language Model (3D-PC-LLM) to mitigate occlusion and viewpoint biases, ensuring more stable and spatially consistent preference signals. Additionally, PRISM leverages Chain-of-Thought (CoT) reasoning to incorporate long-horizon considerations, thereby preventing the short-sighted feedback often seen in static preference comparisons. In contrast to conventional PBRL techniques, this integration of 3D perception and future-oriented reasoning leads to significant gains in preference agreement rates, faster policy convergence, and robust generalization across unseen robotic environments. Our empirical results, spanning tasks such as robotic manipulation and autonomous navigation, highlight PRISM's potential for real-world applications where precise spatial understanding and reliable long-term decision-making are critical. By bridging 3D geometric awareness with CoT-driven preference modeling, PRISM establishes a comprehensive foundation for scalable, human-aligned reinforcement learning.

★ "Well, Keep Thinking": Enhancing LLM Reasoning with Adaptive Injection Decoding

Hyunbin Jin, Je Won Yeom, Seunghyun Bae, Taesup Kim

Large language models (LLMs) exhibit strong reasoning abilities, often attributed to few-shot or zero-shot chain-of-thought (CoT) prompting. While effective, these methods require labor-intensive prompt engineering, raising the question of whether reasoning can be induced without reliance on explicit prompts. In this work, we unlock the reasoning capabilities of LLMs without explicit prompting. Inspired by zero-shot CoT and CoT-decoding, we propose a novel decoding strategy that systematically nudges LLMs to continue reasoning, thereby preventing immature reasoning processes. Specifically, we monitor the model's generation and inject a designated phrase whenever it is likely to conclude its response prematurely, before completing the reasoning process. Our experimental evaluations on diverse reasoning benchmarks demonstrate that our proposed strategy substantially improves LLM reasoning capabilities, highlighting the potential of decoding-based interventions as an alternative to traditional prompting techniques.

★ Retrieval-Augmented Generation with Hierarchical Knowledge

Haoyu Huang, Yongfeng Huang, Junjie Yang, Zhenyu Pan, Yongqiang Chen, Kaili Ma, Hongzhi Chen, James Cheng

Graph-based Retrieval-Augmented Generation (RAG) methods have significantly enhanced the performance of large language models (LLMs) in domain-specific tasks. However, existing RAG methods do not adequately utilize the naturally inherent hierarchical knowledge in human cognition, which limits the capabilities of RAG systems. In this paper, we introduce a new RAG approach, called HiRAG, which utilizes hierarchical knowledge to enhance the semantic understanding and structure capturing capabilities of RAG systems in the indexing and retrieval processes. Our extensive experiments demonstrate that HiRAG achieves significant performance improvements over the state-of-the-art baseline methods. The code of our proposed method is available at \href{https://github.com/hhy-huang/HiRAG}{https://github.com/hhy-huang/HiRAG}.

★ Gumiho: A Hybrid Architecture to Prioritize Early Tokens in Speculative Decoding

Jinze Li, Yixing Xu, Haiduo Huang, Xuanwu Yin, Dong Li, Edith C. H. Ngai, Emad Barsoum

Speculative decoding (SPD) aims to accelerate the auto-regressive token generation process of a target Large Language Model (LLM). Some approaches employ a draft model with multiple heads to predict a sequence of future tokens, where each head handles a token in the sequence. The target LLM verifies the predicted sequence and accepts aligned tokens, enabling efficient multi-token generation. However, existing methods assume that all tokens within a sequence are equally important, employing identical head structures and relying on a single-generation paradigm, either serial or parallel. To this end, we theoretically demonstrate that initial tokens in the draft sequence are more important than later ones. Building on this insight, we propose Gumiho, a hybrid model combining serial and parallel heads. Specifically, given the critical importance of early tokens, we employ a sophisticated Transformer architecture for the early draft heads in a serial configuration to improve accuracy. For later tokens, we utilize multiple lightweight MLP heads operating in parallel to enhance efficiency. By allocating more advanced model structures and longer running times to the early heads, Gumiho achieves improved overall performance. The experimental results demonstrate that our method outperforms existing approaches, fully validating its effectiveness.

comment: Paper under review

★ Cognitive-Mental-LLM: Leveraging Reasoning in Large Language Models for Mental Health Prediction via Online Text

Avinash Patil, Amardeep Kour Gedhu

Large Language Models (LLMs) have demonstrated potential in predicting mental health outcomes from online text, yet traditional classification methods often lack interpretability and robustness. This study evaluates structured reasoning techniques-Chain-of-Thought (CoT), Self-Consistency (SC-CoT), and Tree-of-Thought (ToT)-to improve classification accuracy across multiple mental health datasets sourced from Reddit. We analyze reasoning-driven prompting strategies, including Zero-shot CoT and Few-shot CoT, using key performance metrics such as Balanced Accuracy, F1 score, and Sensitivity/Specificity. Our findings indicate that reasoning-enhanced techniques improve classification performance over direct prediction, particularly in complex cases. Compared to baselines such as Zero Shot non-CoT Prompting, and fine-tuned pre-trained transformers such as BERT and Mental-RoBerta, and fine-tuned Open Source LLMs such as Mental Alpaca and Mental-Flan-T5, reasoning-driven LLMs yield notable gains on datasets like Dreaddit (+0.52\% over M-LLM, +0.82\% over BERT) and SDCNL (+4.67\% over M-LLM, +2.17\% over BERT). However, performance declines in Depression Severity, and CSSRS predictions suggest dataset-specific limitations, likely due to our using a more extensive test set. Among prompting strategies, Few-shot CoT consistently outperforms others, reinforcing the effectiveness of reasoning-driven LLMs. Nonetheless, dataset variability highlights challenges in model reliability and interpretability. This study provides a comprehensive benchmark of reasoning-based LLM techniques for mental health text classification. It offers insights into their potential for scalable clinical applications while identifying key challenges for future improvements.

comment: 8 pages, 4 Figures, 3 tables

★ Semantic Synergy: Unlocking Policy Insights and Learning Pathways Through Advanced Skill Mapping

Phoebe Koundouri, Conrad Landis, Georgios Feretzakis

This research introduces a comprehensive system based on state-of-the-art natural language processing, semantic embedding, and efficient search techniques for retrieving similarities and thus generating actionable insights from raw textual information. The system automatically extracts and aggregates normalized competencies from multiple documents (such as policy files and curricula vitae) and creates strong relationships between recognized competencies, occupation profiles, and related learning courses. To validate its performance, we conducted a multi-tier evaluation that included both explicit and implicit skill references in synthetic and real-world documents. The results showed near-human-level accuracy, with F1 scores exceeding 0.95 for explicit skill detection and above 0.93 for implicit mentions. The system thereby establishes a sound foundation for supporting in-depth collaboration across the AE4RIA network. The methodology involves a multi-stage pipeline based on extensive preprocessing and data cleaning, semantic embedding and segmentation via SentenceTransformer, and skill extraction using a FAISS-based search method. The extracted skills are associated with occupation frameworks (as formulated in the ESCO ontology) and with learning paths offered through the Sustainable Development Goals Academy. Moreover, interactive visualization software, implemented with Dash and Plotly, presents graphs and tables for real-time exploration and informed decision-making by those involved in policymaking, training and learning supply, career transitions, and recruitment. Overall, this system, backed by rigorous validation, offers promising prospects for improved policymaking, human resource development, and lifelong learning by providing structured and actionable insights from raw, complex textual information.

★ Representation-based Reward Modeling for Efficient Safety Alignment of Large Language Model

Qiyuan Deng, Xuefeng Bai, Kehai Chen, Yaowei Wang, Liqiang Nie, Min Zhang

Reinforcement Learning (RL) algorithms for safety alignment of Large Language Models (LLMs), such as Direct Preference Optimization (DPO), encounter the challenge of distribution shift. Current approaches typically address this issue through online sampling from the target policy, which requires significant computational resources. In this paper, we hypothesize that during off-policy training, while the ranking order of output generated by policy changes, their overall distribution remains relatively stable. This stability allows the transformation of the sampling process from the target policy into a re-ranking of preference data. Building on this hypothesis, We propose a new framework that leverages the model's intrinsic safety judgment capability to extract reward signals, which are then used to calculate label confidence for preferences reordering. Extensive experimental results and theoretical analysis demonstrate that the proposed method effectively addresses the distribution shift issue, remarkably enhancing the safety performance while reducing about 300x computational overheads.

★ Why Does Your CoT Prompt (Not) Work? Theoretical Analysis of Prompt Space Complexity, its Interaction with Answer Space During CoT Reasoning with LLMs: A Recurrent Perspective

Xiang Zhang, Juntai Cao, Jiaqi Wei, Chenyu You, Dujian Ding

Despite the remarkable successes of Large Language Models (LLMs), their fundamental Transformer architecture possesses inherent theoretical limitations that restrict their capability to handle reasoning tasks with increasing computational complexity. Chain-of-Thought (CoT) prompting has emerged as a practical solution, supported by several theoretical studies. However, current CoT-based methods (including ToT, GoT, etc.) generally adopt a "one-prompt-fits-all" strategy, using fixed templates (e.g., "think step by step") across diverse reasoning tasks. This method forces models to navigate an extremely complex prompt space to identify effective reasoning paths. The current prompt designing research are also heavily relying on trial-and-error rather than theoretically informed guidance. In this paper, we provide a rigorous theoretical analysis of the complexity and interplay between two crucial spaces: the prompt space (the space of potential prompt structures) and the answer space (the space of reasoning solutions generated by LLMs) in CoT reasoning. We demonstrate how reliance on a single universal prompt (e.g. think step by step) can negatively impact the theoretical computability of LLMs, illustrating that prompt complexity directly influences the structure and effectiveness of the navigation in answer space. Our analysis highlights that sometimes human supervision is critical for efficiently navigating the prompt space. We theoretically and empirically show that task-specific prompting significantly outperforms unsupervised prompt generation, emphasizing the necessity of thoughtful human guidance in CoT prompting.

comment: arXiv admin note: substantial text overlap with arXiv:2410.14198

★ Information Density Principle for MLLM Benchmarks

Chunyi Li, Xiaozhe Li, Zicheng Zhang, Yuan Tian, Ziheng Jia, Xiaohong Liu, Xiongkuo Min, Jia Wang, Haodong Duan, Kai Chen, Guangtao Zhai

With the emergence of Multimodal Large Language Models (MLLMs), hundreds of benchmarks have been developed to ensure the reliability of MLLMs in downstream tasks. However, the evaluation mechanism itself may not be reliable. For developers of MLLMs, questions remain about which benchmark to use and whether the test results meet their requirements. Therefore, we propose a critical principle of Information Density, which examines how much insight a benchmark can provide for the development of MLLMs. We characterize it from four key dimensions: (1) Fallacy, (2) Difficulty, (3) Redundancy, (4) Diversity. Through a comprehensive analysis of more than 10,000 samples, we measured the information density of 19 MLLM benchmarks. Experiments show that using the latest benchmarks in testing can provide more insight compared to previous ones, but there is still room for improvement in their information density. We hope this principle can promote the development and application of future MLLM benchmarks. Project page: https://github.com/lcysyzxdxc/bench4bench

★ Compute Optimal Scaling of Skills: Knowledge vs Reasoning

Nicholas Roberts, Niladri Chatterji, Sharan Narang, Mike Lewis, Dieuwke Hupkes

Scaling laws are a critical component of the LLM development pipeline, most famously as a way to forecast training decisions such as 'compute-optimally' trading-off parameter count and dataset size, alongside a more recent growing list of other crucial decisions. In this work, we ask whether compute-optimal scaling behaviour can be skill-dependent. In particular, we examine knowledge and reasoning-based skills such as knowledge-based QA and code generation, and we answer this question in the affirmative: $\textbf{scaling laws are skill-dependent}$. Next, to understand whether skill-dependent scaling is an artefact of the pretraining datamix, we conduct an extensive ablation of different datamixes and find that, also when correcting for datamix differences, $\textbf{knowledge and code exhibit fundamental differences in scaling behaviour}$. We conclude with an analysis of how our findings relate to standard compute-optimal scaling using a validation set, and find that $\textbf{a misspecified validation set can impact compute-optimal parameter count by nearly 50%,}$ depending on its skill composition.

★ Using Context to Improve Word Segmentation

Stephanie Hu, Xiaolu Guo

An important step in understanding how children acquire languages is studying how infants learn word segmentation. It has been established in previous research that infants may use statistical regularities in speech to learn word segmentation. The research of Goldwater et al., demonstrated that incorporating context in models improves their ability to learn word segmentation. We implemented two of their models, a unigram and bigram model, to examine how context can improve statistical word segmentation. The results are consistent with our hypothesis that the bigram model outperforms the unigram model at predicting word segmentation. Extending the work of Goldwater et al., we also explored basic ways to model how young children might use previously learned words to segment new utterances.

★ ExtremeAIGC: Benchmarking LMM Vulnerability to AI-Generated Extremist Content

Bhavik Chandna, Mariam Aboujenane, Usman Naseem

Large Multimodal Models (LMMs) are increasingly vulnerable to AI-generated extremist content, including photorealistic images and text, which can be used to bypass safety mechanisms and generate harmful outputs. However, existing datasets for evaluating LMM robustness offer limited exploration of extremist content, often lacking AI-generated images, diverse image generation models, and comprehensive coverage of historical events, which hinders a complete assessment of model vulnerabilities. To fill this gap, we introduce ExtremeAIGC, a benchmark dataset and evaluation framework designed to assess LMM vulnerabilities against such content. ExtremeAIGC simulates real-world events and malicious use cases by curating diverse text- and image-based examples crafted using state-of-the-art image generation techniques. Our study reveals alarming weaknesses in LMMs, demonstrating that even cutting-edge safety measures fail to prevent the generation of extremist material. We systematically quantify the success rates of various attack strategies, exposing critical gaps in current defenses and emphasizing the need for more robust mitigation strategies.

comment: Preprint

★ Take Off the Training Wheels Progressive In-Context Learning for Effective Alignment EMNLP2024

Zhenyu Liu, Dongfang Li, Xinshuo Hu, Xinping Zhao, Yibin Chen, Baotian Hu, Min Zhang

Recent studies have explored the working mechanisms of In-Context Learning (ICL). However, they mainly focus on classification and simple generation tasks, limiting their broader application to more complex generation tasks in practice. To address this gap, we investigate the impact of demonstrations on token representations within the practical alignment tasks. We find that the transformer embeds the task function learned from demonstrations into the separator token representation, which plays an important role in the generation of prior response tokens. Once the prior response tokens are determined, the demonstrations become redundant.Motivated by this finding, we propose an efficient Progressive In-Context Alignment (PICA) method consisting of two stages. In the first few-shot stage, the model generates several prior response tokens via standard ICL while concurrently extracting the ICL vector that stores the task function from the separator token representation. In the following zero-shot stage, this ICL vector guides the model to generate responses without further demonstrations.Extensive experiments demonstrate that our PICA not only surpasses vanilla ICL but also achieves comparable performance to other alignment tuning methods. The proposed training-free method reduces the time cost (e.g., 5.45+) with improved alignment performance (e.g., 6.57+). Consequently, our work highlights the application of ICL for alignment and calls for a deeper understanding of ICL for complex generations. The code will be available at https://github.com/HITsz-TMG/PICA.

comment: 15 pages, 9 figures, published in EMNLP2024

★ Developing and Evaluating an AI-Assisted Prediction Model for Unplanned Intensive Care Admissions following Elective Neurosurgery using Natural Language Processing within an Electronic Healthcare Record System

Julia Ive, Olatomiwa Olukoya, Jonathan P. Funnell, James Booker, Sze H M Lam, Ugan Reddy, Kawsar Noor, Richard JB Dobson, Astri M. V. Luoma, Hani J Marcus

Introduction: Timely care in a specialised neuro-intensive therapy unit (ITU) reduces mortality and hospital stays, with planned admissions being safer than unplanned ones. However, post-operative care decisions remain subjective. This study used artificial intelligence (AI), specifically natural language processing (NLP) to analyse electronic health records (EHRs) and predict ITU admissions for elective surgery patients. Methods: This study analysed the EHRs of elective neurosurgery patients from University College London Hospital (UCLH) using NLP. Patients were categorised into planned high dependency unit (HDU) or ITU admission; unplanned HDU or ITU admission; or ward / overnight recovery (ONR). The Medical Concept Annotation Tool (MedCAT) was used to identify SNOMED-CT concepts within the clinical notes. We then explored the utility of these identified concepts for a range of AI algorithms trained to predict ITU admission. Results: The CogStack-MedCAT NLP model, initially trained on hospital-wide EHRs, underwent two refinements: first with data from patients with Normal Pressure Hydrocephalus (NPH) and then with data from Vestibular Schwannoma (VS) patients, achieving a concept detection F1-score of 0.93. This refined model was then used to extract concepts from EHR notes of 2,268 eligible neurosurgical patients. We integrated the extracted concepts into AI models, including a decision tree model and a neural time-series model. Using the simpler decision tree model, we achieved a recall of 0.87 (CI 0.82 - 0.91) for ITU admissions, reducing the proportion of unplanned ITU cases missed by human experts from 36% to 4%. Conclusion: The NLP model, refined for accuracy, has proven its efficiency in extracting relevant concepts, providing a reliable basis for predictive AI models to use in clinically valid applications.

★ PluralLLM: Pluralistic Alignment in LLMs via Federated Learning

Mahmoud Srewa, Tianyu Zhao, Salma Elmalaki

Ensuring Large Language Models (LLMs) align with diverse human preferences while preserving privacy and fairness remains a challenge. Existing methods, such as Reinforcement Learning from Human Feedback (RLHF), rely on centralized data collection, making them computationally expensive and privacy-invasive. We introduce PluralLLM a federated learning-based approach that enables multiple user groups to collaboratively train a transformer-based preference predictor without sharing sensitive data, which can also serve as a reward model for aligning LLMs. Our method leverages Federated Averaging (FedAvg) to aggregate preference updates efficiently, achieving 46% faster convergence, a 4% improvement in alignment scores, and nearly the same group fairness measure as in centralized training. Evaluated on a Q/A preference alignment task, PluralLLM demonstrates that federated preference learning offers a scalable and privacy-preserving alternative for aligning LLMs with diverse human values.

♻ ★ Chain-of-Thought Reasoning In The Wild Is Not Always Faithful ICLR 25

Iván Arcuschin, Jett Janiak, Robert Krzyzanowski, Senthooran Rajamanoharan, Neel Nanda, Arthur Conmy

Chain-of-Thought (CoT) reasoning has significantly advanced state-of-the-art AI capabilities. However, recent studies have shown that CoT reasoning is not always faithful, i.e. CoT reasoning does not always reflect how models arrive at conclusions. So far, most of these studies have focused on unfaithfulness in unnatural contexts where an explicit bias has been introduced. In contrast, we show that unfaithful CoT can occur on realistic prompts with no artificial bias. Our results reveal non-negligible rates of several forms of unfaithful reasoning in frontier models: Sonnet 3.7 (16.3%), DeepSeek R1 (5.3%) and ChatGPT-4o (7.0%) all answer a notable proportion of question pairs unfaithfully. Specifically, we find that models rationalize their implicit biases in answers to binary questions ("implicit post-hoc rationalization"). For example, when separately presented with the questions "Is X bigger than Y?" and "Is Y bigger than X?", models sometimes produce superficially coherent arguments to justify answering Yes to both questions or No to both questions, despite such responses being logically contradictory. We also investigate restoration errors (Dziri et al., 2023), where models make and then silently correct errors in their reasoning, and unfaithful shortcuts, where models use clearly illogical reasoning to simplify solving problems in Putnam questions (a hard benchmark). Our findings raise challenges for AI safety work that relies on monitoring CoT to detect undesired behavior.

comment: Accepted to the Reasoning and Planning for Large Language Models Workshop (ICLR 25), 10 main paper pages, 38 appendix pages

♻ ★ DataEnvGym: Data Generation Agents in Teacher Environments with Student Feedback ICLR 2025

Zaid Khan, Elias Stengel-Eskin, Jaemin Cho, Mohit Bansal

The process of creating training data to teach models is currently driven by humans, who manually analyze model weaknesses and plan how to create data that improves a student model. Approaches using LLMs as annotators reduce human effort, but still require humans to interpret feedback from evaluations and control the LLM to produce data the student needs. Automating this labor-intensive process by creating autonomous data generation agents - or teachers - is desirable, but requires environments that can simulate the feedback-driven, iterative, closed loop of data creation. To enable rapid, scalable testing for such agents and their modules, we introduce DataEnvGym, a testbed of teacher environments for data generation agents. DataEnvGym frames data generation as a sequential decision-making task, involving an agent consisting of a data generation policy (which generates a plan for creating training data) and a data generation engine (which transforms the plan into data), inside an environment that provides student feedback. The agent's goal is to improve student performance. Students are iteratively trained and evaluated on generated data, and their feedback (in the form of errors or weak skills) is reported to the agent after each iteration. DataEnvGym includes multiple teacher environment instantiations across 3 levels of structure in the state representation and action space. More structured environments are based on inferred skills and offer more interpretability and curriculum control. We support 4 domains (math, code, VQA, and tool-use) and test multiple students and teachers. Example agents in our teaching environments can iteratively improve students across tasks and settings. Moreover, we show that environments teach different skill levels and test variants of key modules, pointing to future work in improving data generation agents, engines, and feedback mechanisms.

comment: ICLR 2025 Spotlight; Project Page: https://DataEnvGym.github.io

♻ ★ YouTube Comments Decoded: Leveraging LLMs for Low Resource Language Classification

Aniket Deroy, Subhankar Maity

Sarcasm detection is a significant challenge in sentiment analysis, particularly due to its nature of conveying opinions where the intended meaning deviates from the literal expression. This challenge is heightened in social media contexts where code-mixing, especially in Dravidian languages, is prevalent. Code-mixing involves the blending of multiple languages within a single utterance, often with non-native scripts, complicating the task for systems trained on monolingual data. This shared task introduces a novel gold standard corpus designed for sarcasm and sentiment detection within code-mixed texts, specifically in Tamil-English and Malayalam-English languages. The primary objective of this task is to identify sarcasm and sentiment polarity within a code-mixed dataset of Tamil-English and Malayalam-English comments and posts collected from social media platforms. Each comment or post is annotated at the message level for sentiment polarity, with particular attention to the challenges posed by class imbalance, reflecting real-world scenarios.In this work, we experiment with state-of-the-art large language models like GPT-3.5 Turbo via prompting to classify comments into sarcastic or non-sarcastic categories. We obtained a macro-F1 score of 0.61 for Tamil language. We obtained a macro-F1 score of 0.50 for Malayalam language.

comment: Updated and Final Version

♻ ★ Joint Fine-tuning and Conversion of Pretrained Speech and Language Models towards Linear Complexity ICLR 2025

Mutian He, Philip N. Garner

Architectures such as Linformer and Mamba have recently emerged as competitive linear time replacements for transformers. However, corresponding large pretrained models are often unavailable, especially in non-text domains. To remedy this, we present a Cross-Architecture Layerwise Distillation (CALD) approach that jointly converts a transformer model to a linear time substitute and fine-tunes it to a target task. We also compare several means to guide the fine-tuning to optimally retain the desired inference capability from the original model. The methods differ in their use of the target model and the trajectory of the parameters. In a series of empirical studies on language processing, language modeling, and speech processing, we show that CALD can effectively recover the result of the original model, and that the guiding strategy contributes to the result. Some reasons for the variation are suggested.

comment: 18 pages, 5 figures; ICLR 2025 camera ready. Code: https://github.com/idiap/linearize-distill-pretrained-transformers

♻ ★ Latent Space Chain-of-Embedding Enables Output-free LLM Self-Evaluation ICLR 2025

Yiming Wang, Pei Zhang, Baosong Yang, Derek F. Wong, Rui Wang

LLM self-evaluation relies on the LLM's own ability to estimate response correctness, which can greatly improve its deployment reliability. In this research track, we propose the Chain-of-Embedding (CoE) in the latent space to enable LLMs to perform output-free self-evaluation. CoE consists of all progressive hidden states produced during the inference time, which can be treated as the latent thinking path of LLMs. We find that when LLMs respond correctly and incorrectly, their CoE features differ, these discrepancies assist us in estimating LLM response correctness. Experiments in four diverse domains and seven LLMs fully demonstrate the effectiveness of our method. Meanwhile, its label-free design intent without any training and millisecond-level computational cost ensures real-time feedback in large-scale scenarios. More importantly, we provide interesting insights into LLM response correctness from the perspective of hidden state changes inside LLMs.

comment: Accepted by ICLR 2025

♻ ★ When Text Embedding Meets Large Language Model: A Comprehensive Survey

Zhijie Nie, Zhangchi Feng, Mingxin Li, Cunwang Zhang, Yanzhao Zhang, Dingkun Long, Richong Zhang

Text embedding has become a foundational technology in natural language processing (NLP) during the deep learning era, driving advancements across a wide array of downstream tasks. While many natural language understanding challenges can now be modeled using generative paradigms and leverage the robust generative and comprehension capabilities of large language models (LLMs), numerous practical applications-such as semantic matching, clustering, and information retrieval-continue to rely on text embeddings for their efficiency and effectiveness. Therefore, how to combine the LLMs and the text embeddings has become one of the hotspots of academic attention in recent years. In this survey, we categorize the interplay between LLMs and text embeddings into three overarching themes: (1) LLM-augmented text embedding, enhancing traditional embedding methods with LLMs; (2) LLMs as text embedders, adapting their innate capabilities for high-quality embedding; and (3) Text embedding understanding with LLMs, leveraging LLMs to analyze and interpret embeddings. By organizing recent works based on interaction patterns rather than specific downstream applications, we offer a novel and systematic overview of contributions from various research and application domains in the era of LLMs. Furthermore, we highlight the unresolved challenges that persisted in the pre-LLM era with pre-trained language models (PLMs) and explore the emerging obstacles brought forth by LLMs. Building on this analysis, we outline prospective directions for the evolution of text embedding, addressing both theoretical and practical opportunities in the rapidly advancing landscape of NLP.

comment: Work in progress

♻ ★ InftyThink: Breaking the Length Limits of Long-Context Reasoning in Large Language Models

Yuchen Yan, Yongliang Shen, Yang Liu, Jin Jiang, Mengdi Zhang, Jian Shao, Yueting Zhuang

Advanced reasoning in large language models has achieved remarkable performance on challenging tasks, but the prevailing long-context reasoning paradigm faces critical limitations: quadratic computational scaling with sequence length, reasoning constrained by maximum context boundaries, and performance degradation beyond pre-training context windows. Existing approaches primarily compress reasoning chains without addressing the fundamental scaling problem. To overcome these challenges, we introduce InftyThink, a paradigm that transforms monolithic reasoning into an iterative process with intermediate summarization. By interleaving short reasoning segments with concise progress summaries, our approach enables unbounded reasoning depth while maintaining bounded computational costs. This creates a characteristic sawtooth memory pattern that significantly reduces computational complexity compared to traditional approaches. Furthermore, we develop a methodology for reconstructing long-context reasoning datasets into our iterative format, transforming OpenR1-Math into 333K training instances. Experiments across multiple model architectures demonstrate that our approach reduces computational costs while improving performance, with Qwen2.5-Math-7B showing 3-13% improvements across MATH500, AIME24, and GPQA_diamond benchmarks. Our work challenges the assumed trade-off between reasoning depth and computational efficiency, providing a more scalable approach to complex reasoning without architectural modifications.

♻ ★ DataMan: Data Manager for Pre-training Large Language Models ICLR2025

Ru Peng, Kexin Yang, Yawen Zeng, Junyang Lin, Dayiheng Liu, Junbo Zhao

The performance emergence of large language models (LLMs) driven by data scaling laws makes the selection of pre-training data increasingly important. However, existing methods rely on limited heuristics and human intuition, lacking comprehensive and clear guidelines. To address this, we are inspired by ``reverse thinking'' -- prompting LLMs to self-identify which criteria benefit its performance. As its pre-training capabilities are related to perplexity (PPL), we derive 14 quality criteria from the causes of text perplexity anomalies and introduce 15 common application domains to support domain mixing. In this paper, we train a Data Manager (DataMan) to learn quality ratings and domain recognition from pointwise rating, and use it to annotate a 447B token pre-training corpus with 14 quality ratings and domain type. Our experiments validate our approach, using DataMan to select 30B tokens to train a 1.3B-parameter language model, demonstrating significant improvements in in-context learning (ICL), perplexity, and instruction-following ability over the state-of-the-art baseline. The best-performing model, based on the Overall Score l=5 surpasses a model trained with 50% more data using uniform sampling. We continue pre-training with high-rated, domain-specific data annotated by DataMan to enhance domain-specific ICL performance and thus verify DataMan's domain mixing ability. Our findings emphasize the importance of quality ranking, the complementary nature of quality criteria, and their low correlation with perplexity, analyzing misalignment between PPL and ICL performance. We also thoroughly analyzed our pre-training dataset, examining its composition, the distribution of quality ratings, and the original document sources.

comment: ICLR2025 paper

♻ ★ MastermindEval: A Simple But Scalable Reasoning Benchmark ICLR 2025

Jonas Golde, Patrick Haller, Fabio Barth, Alan Akbik

Recent advancements in large language models (LLMs) have led to remarkable performance across a wide range of language understanding and mathematical tasks. As a result, increasing attention has been given to assessing the true reasoning capabilities of LLMs, driving research into commonsense, numerical, logical, and qualitative reasoning. However, with the rapid progress of reasoning-focused models such as OpenAI's o1 and DeepSeek's R1, there has been a growing demand for reasoning benchmarks that can keep pace with ongoing model developments. In this paper, we introduce MastermindEval, a simple, scalable, and interpretable deductive reasoning benchmark inspired by the board game Mastermind. Our benchmark supports two evaluation paradigms: (1) agentic evaluation, in which the model autonomously plays the game, and (2) deductive reasoning evaluation, in which the model is given a pre-played game state with only one possible valid code to infer. In our experimental results we (1) find that even easy Mastermind instances are difficult for current models and (2) demonstrate that the benchmark is scalable to possibly more advanced models in the future Furthermore, we investigate possible reasons why models cannot deduce the final solution and find that current models are limited in deducing the concealed code as the number of statement to combine information from is increasing.

comment: 9 pages, 2 figures, 4 tables. In: ICLR 2025 Workshop on Reasoning and Planning for Large Language Models

♻ ★ EMOVA: Empowering Language Models to See, Hear and Speak with Vivid Emotions CVPR 2025

Kai Chen, Yunhao Gou, Runhui Huang, Zhili Liu, Daxin Tan, Jing Xu, Chunwei Wang, Yi Zhu, Yihan Zeng, Kuo Yang, Dingdong Wang, Kun Xiang, Haoyuan Li, Haoli Bai, Jianhua Han, Xiaohui Li, Weike Jin, Nian Xie, Yu Zhang, James T. Kwok, Hengshuang Zhao, Xiaodan Liang, Dit-Yan Yeung, Xiao Chen, Zhenguo Li, Wei Zhang, Qun Liu, Jun Yao, Lanqing Hong, Lu Hou, Hang Xu

GPT-4o, an omni-modal model that enables vocal conversations with diverse emotions and tones, marks a milestone for omni-modal foundation models. However, empowering Large Language Models to perceive and generate images, texts, and speeches end-to-end with publicly available data remains challenging for the open-source community. Existing vision-language models rely on external tools for speech processing, while speech-language models still suffer from limited or totally without vision-understanding capabilities. To address this gap, we propose the EMOVA (EMotionally Omni-present Voice Assistant), to enable Large Language Models with end-to-end speech abilities while maintaining the leading vision-language performance. With a semantic-acoustic disentangled speech tokenizer, we surprisingly notice that omni-modal alignment can further enhance vision-language and speech abilities compared with the bi-modal aligned counterparts. Moreover, a lightweight style module is introduced for the flexible speech style controls including emotions and pitches. For the first time, EMOVA achieves state-of-the-art performance on both the vision-language and speech benchmarks, and meanwhile, supporting omni-modal spoken dialogue with vivid emotions.

comment: Accepted by CVPR 2025. Project Page: https://emova-ollm.github.io/

♻ ★ MIX : a Multi-task Learning Approach to Solve Open-Domain Question Answering

Sofian Chaybouti, Achraf Saghe, Aymen Shabou

This paper introduces MIX, a multi-task deep learning approach to solve open-ended question-answering. First, we design our system as a multi-stage pipeline of 3 building blocks: a BM25-based Retriever to reduce the search space, a RoBERTa-based Scorer, and an Extractor to rank retrieved paragraphs and extract relevant text spans, respectively. Eventually, we further improve the computational efficiency of our system to deal with the scalability challenge: thanks to multi-task learning, we parallelize the close tasks solved by the Scorer and the Extractor. Our system is on par with state-of-the-art performances on the squad-open benchmark while being simpler conceptually.

comment: 8 pages, 7 figures, 3 tables

♻ ★ PAD: Personalized Alignment of LLMs at Decoding-Time ICLR 2025

Ruizhe Chen, Xiaotian Zhang, Meng Luo, Wenhao Chai, Zuozhu Liu

Aligning with personalized preferences, which vary significantly across cultural, educational, and political differences, poses a significant challenge due to the computational costs and data demands of traditional alignment methods. In response, this paper presents Personalized Alignment at Decoding-time (PAD), a novel framework designed to align LLM outputs with diverse personalized preferences during the inference phase, eliminating the need for additional training. By introducing a unique personalized reward modeling strategy, this framework decouples the text generation process from personalized preferences, facilitating the generation of generalizable token-level personalized rewards. The PAD algorithm leverages these rewards to guide the decoding process, dynamically tailoring the base model's predictions to personalized preferences. Extensive experimental results demonstrate that PAD not only outperforms existing training-based alignment methods in terms of aligning with diverse preferences but also shows significant generalizability to preferences unseen during training and scalability across different base models. This work advances the capability of LLMs to meet user needs in real-time applications, presenting a substantial step forward in personalized LLM alignment.

comment: ICLR 2025

♻ ★ Diabetica: Adapting Large Language Model to Enhance Multiple Medical Tasks in Diabetes Care and Management ICLR 2025

Lai Wei, Zhen Ying, Muyang He, Yutong Chen, Qian Yang, Yanzhe Hong, Jiaping Lu, Kaipeng Zheng, Shaoting Zhang, Xiaoying Li, Weiran Huang, Ying Chen

Diabetes is a chronic disease with a significant global health burden, requiring multi-stakeholder collaboration for optimal management. Large language models (LLMs) have shown promise in various healthcare scenarios, but their effectiveness across diverse diabetes tasks remains unproven. Our study introduced a framework to train and validate diabetes-specific LLMs. We first developed a comprehensive data processing pipeline that includes data collection, filtering, augmentation and refinement. This created a high-quality, diabetes-specific dataset and evaluation benchmarks from scratch. Fine-tuned on the collected training dataset, our diabetes-specific LLM family demonstrated state-of-the-art proficiency in processing various diabetes tasks compared to other LLMs. Furthermore, clinical studies revealed the potential applications of our models in diabetes care, including providing personalized healthcare, assisting medical education, and streamlining clinical tasks. Generally, our introduced framework helps develop diabetes-specific LLMs and highlights their potential to enhance clinical practice and provide personalized, data-driven support for diabetes management across different end users. Our codes, benchmarks and models are available at https://github.com/waltonfuture/Diabetica.

comment: Accepted by ICLR 2025 SCI-FM workshop

♻ ★ Adapting Multilingual Embedding Models to Historical Luxembourgish

Andrianos Michail, Corina Julia Raclé, Juri Opitz, Simon Clematide

The growing volume of digitized historical texts requires effective semantic search using text embeddings. However, pre-trained multilingual models face challenges with historical content due to OCR noise and outdated spellings. This study examines multilingual embeddings for cross-lingual semantic search in historical Luxembourgish (LB), a low-resource language. We collect historical Luxembourgish news articles from various periods and use GPT-4o for sentence segmentation and translation, generating 20,000 parallel training sentences per language pair. Additionally, we create a semantic search (Historical LB Bitext Mining) evaluation set and find that existing models perform poorly on cross-lingual search for historical Luxembourgish. Using our historical and additional modern parallel training data, we adapt several multilingual embedding models through contrastive learning or knowledge distillation and increase accuracy significantly for all models. We release our adapted models and historical Luxembourgish-German/French/English bitexts to support further research.

comment: To appear in LaTeCH-CLfL 2025

♻ ★ FIND: Fine-grained Information Density Guided Adaptive Retrieval-Augmented Generation for Disease Diagnosis

Mingyi Jia, Junwen Duan, Yan Song, Jianxin Wang

Retrieval-Augmented Large Language Models (LLMs), which integrate external knowledge into LLMs, have shown remarkable performance in various medical domains, including clinical diagnosis. However, existing RAG methods struggle to effectively assess task difficulty to make retrieval decisions, thereby failing to meet the clinical requirements for balancing efficiency and accuracy. So in this paper, we propose FIND (\textbf{F}ine-grained \textbf{In}formation \textbf{D}ensity Guided Adaptive RAG), a novel framework that improves the reliability of RAG in disease diagnosis scenarios. FIND incorporates a fine-grained adaptive control module to determine whether retrieval is necessary based on the information density of the input. By optimizing the retrieval process and implementing a knowledge filtering module, FIND ensures that the retrieval is better suited to clinical scenarios. Experiments on three Chinese electronic medical record datasets demonstrate that FIND significantly outperforms various baseline methods, highlighting its effectiveness in clinical diagnosis tasks.

♻ ★ Automated Knowledge Concept Annotation and Question Representation Learning for Knowledge Tracing

Yilmazcan Ozyurt, Stefan Feuerriegel, Mrinmaya Sachan

Knowledge tracing (KT) is a popular approach for modeling students' learning progress over time, which can enable more personalized and adaptive learning. However, existing KT approaches face two major limitations: (1) they rely heavily on expert-defined knowledge concepts (KCs) in questions, which is time-consuming and prone to errors; and (2) KT methods tend to overlook the semantics of both questions and the given KCs. In this work, we address these challenges and present KCQRL, a framework for automated knowledge concept annotation and question representation learning that can improve the effectiveness of any existing KT model. First, we propose an automated KC annotation process using large language models (LLMs), which generates question solutions and then annotates KCs in each solution step of the questions. Second, we introduce a contrastive learning approach to generate semantically rich embeddings for questions and solution steps, aligning them with their associated KCs via a tailored false negative elimination approach. These embeddings can be readily integrated into existing KT models, replacing their randomly initialized embeddings. We demonstrate the effectiveness of KCQRL across 15 KT algorithms on two large real-world Math learning datasets, where we achieve consistent performance improvements.

♻ ★ Is My Text in Your AI Model? Gradient-based Membership Inference Test applied to LLMs

Gonzalo Mancera, Daniel DeAlcala, Julian Fierrez, Ruben Tolosana, Aythami Morales

This work adapts and studies the gradient-based Membership Inference Test (gMINT) to the classification of text based on LLMs. MINT is a general approach intended to determine if given data was used for training machine learning models, and this work focuses on its application to the domain of Natural Language Processing. Using gradient-based analysis, the MINT model identifies whether particular data samples were included during the language model training phase, addressing growing concerns about data privacy in machine learning. The method was evaluated in seven Transformer-based models and six datasets comprising over 2.5 million sentences, focusing on text classification tasks. Experimental results demonstrate MINTs robustness, achieving AUC scores between 85% and 99%, depending on data size and model architecture. These findings highlight MINTs potential as a scalable and reliable tool for auditing machine learning models, ensuring transparency, safeguarding sensitive data, and fostering ethical compliance in the deployment of AI/NLP technologies.

♻ ★ DSMoE: Matrix-Partitioned Experts with Dynamic Routing for Computation-Efficient Dense LLMs

Minxuan Lv, Zhenpeng Su, Leiyu Pan, Yizhe Xiong, Zijia Lin, Hui Chen, Wei Zhou, Jungong Han, Guiguang Ding, Cheng Luo, Di Zhang, Kun Gai, Songlin Hu

As large language models continue to scale, computational costs and resource consumption have emerged as significant challenges. While existing sparsification methods like pruning reduce computational overhead, they risk losing model knowledge through parameter removal. This paper proposes DSMoE (Dynamic Sparse Mixture-of-Experts), a novel approach that achieves sparsification by partitioning pre-trained FFN layers into computational blocks. We implement adaptive expert routing using sigmoid activation and straight-through estimators, enabling tokens to flexibly access different aspects of model knowledge based on input complexity. Additionally, we introduce a sparsity loss term to balance performance and computational efficiency. Extensive experiments on LLaMA models demonstrate that under equivalent computational constraints, DSMoE achieves superior performance compared to existing pruning and MoE approaches across language modeling and downstream tasks, particularly excelling in generation tasks. Analysis reveals that DSMoE learns distinctive layerwise activation patterns, providing new insights for future MoE architecture design.

♻ ★ Evaluating LLMs and Pre-trained Models for Text Summarization Across Diverse Datasets

Tohida Rehman, Soumabha Ghosh, Kuntal Das, Souvik Bhattacharjee, Debarshi Kumar Sanyal, Samiran Chattopadhyay

Text summarization plays a crucial role in natural language processing by condensing large volumes of text into concise and coherent summaries. As digital content continues to grow rapidly and the demand for effective information retrieval increases, text summarization has become a focal point of research in recent years. This study offers a thorough evaluation of four leading pre-trained and open-source large language models: BART, FLAN-T5, LLaMA-3-8B, and Gemma-7B, across five diverse datasets CNN/DM, Gigaword, News Summary, XSum, and BBC News. The evaluation employs widely recognized automatic metrics, including ROUGE-1, ROUGE-2, ROUGE-L, BERTScore, and METEOR, to assess the models' capabilities in generating coherent and informative summaries. The results reveal the comparative strengths and limitations of these models in processing various text types.

comment: 5 pages, 2 figures, 6 tables

♻ ★ Next-Generation Database Interfaces: A Survey of LLM-based Text-to-SQL

Zijin Hong, Zheng Yuan, Qinggang Zhang, Hao Chen, Junnan Dong, Feiran Huang, Xiao Huang

Generating accurate SQL from users' natural language questions (text-to-SQL) remains a long-standing challenge due to the complexities involved in user question understanding, database schema comprehension, and SQL generation. Traditional text-to-SQL systems, which combine human engineering and deep neural networks, have made significant progress. Subsequently, pre-trained language models (PLMs) have been developed for text-to-SQL tasks, achieving promising results. However, as modern databases and user questions grow more complex, PLMs with a limited parameter size often produce incorrect SQL. This necessitates more sophisticated and tailored optimization methods, which restricts the application of PLM-based systems. Recently, large language models (LLMs) have shown significant capabilities in natural language understanding as model scale increases. Thus, integrating LLM-based solutions can bring unique opportunities, improvements, and solutions to text-to-SQL research. In this survey, we provide a comprehensive review of existing LLM-based text-to-SQL studies. Specifically, we offer a brief overview of the technical challenges and evolutionary process of text-to-SQL. Next, we introduce the datasets and metrics designed to evaluate text-to-SQL systems. Subsequently, we present a systematic analysis of recent advances in LLM-based text-to-SQL. Finally, we make a summarization and discuss the remaining challenges in this field and suggest expectations for future research directions.

♻ ★ Computational Law: Datasets, Benchmarks, and Ontologies

Dilek Küçük, Fazli Can

Recent developments in computer science and artificial intelligence have also contributed to the legal domain, as revealed by the number and range of related publications and applications. Machine and deep learning models require considerable amount of domain-specific data for training and comparison purposes, in order to attain high-performance in the legal domain. Additionally, semantic resources such as ontologies are valuable for building large-scale computational legal systems, in addition to ensuring interoperability of such systems. Considering these aspects, we present an up-to-date review of the literature on datasets, benchmarks, and ontologies proposed for computational law. We believe that this comprehensive and recent review will help researchers and practitioners when developing and testing approaches and systems for computational law.

♻ ★ TPO: Aligning Large Language Models with Multi-branch & Multi-step Preference Trees

Weibin Liao, Xu Chu, Yasha Wang

In the domain of complex reasoning tasks, such as mathematical reasoning, recent advancements have proposed the use of Direct Preference Optimization (DPO) to suppress output of dispreferred responses, thereby enhancing the long-chain reasoning capabilities of large language models (LLMs). To this end, these studies employed LLMs to generate preference trees via Tree-of-thoughts (ToT) and sample the paired preference responses required by the DPO algorithm. However, the DPO algorithm based on binary preference optimization is unable to learn multiple responses with varying degrees of preference/dispreference that provided by the preference trees, resulting in incomplete preference learning. In this work, we introduce Tree Preference Optimization (TPO), that does not sample paired preference responses from the preference tree; instead, it directly learns from the entire preference tree during the fine-tuning. Specifically, TPO formulates the language model alignment as a Preference List Ranking problem, where the policy can potentially learn more effectively from a ranked preference list of responses given the prompt. In addition, to further assist LLMs in identifying discriminative steps within long-chain reasoning and increase the relative reward margin in the preference list, TPO utilizes Adaptive Step Reward to adjust the reward values of each step in trajectory for performing fine-grained preference optimization. We carry out extensive experiments on mathematical reasoning tasks to evaluate TPO. The experimental results indicate that TPO consistently outperforms DPO across five public large language models on four datasets.

♻ ★ Prompt-Driven Contrastive Learning for Transferable Adversarial Attacks ECCV 2024

Hunmin Yang, Jongoh Jeong, Kuk-Jin Yoon

Recent vision-language foundation models, such as CLIP, have demonstrated superior capabilities in learning representations that can be transferable across diverse range of downstream tasks and domains. With the emergence of such powerful models, it has become crucial to effectively leverage their capabilities in tackling challenging vision tasks. On the other hand, only a few works have focused on devising adversarial examples that transfer well to both unknown domains and model architectures. In this paper, we propose a novel transfer attack method called PDCL-Attack, which leverages the CLIP model to enhance the transferability of adversarial perturbations generated by a generative model-based attack framework. Specifically, we formulate an effective prompt-driven feature guidance by harnessing the semantic representation power of text, particularly from the ground-truth class labels of input images. To the best of our knowledge, we are the first to introduce prompt learning to enhance the transferable generative attacks. Extensive experiments conducted across various cross-domain and cross-model settings empirically validate our approach, demonstrating its superiority over state-of-the-art methods.

comment: Accepted to ECCV 2024 (Oral), Project Page: https://PDCL-Attack.github.io

♻ ★ Mono-InternVL: Pushing the Boundaries of Monolithic Multimodal Large Language Models with Endogenous Visual Pre-training CVPR 2025

Gen Luo, Xue Yang, Wenhan Dou, Zhaokai Wang, Jiawen Liu, Jifeng Dai, Yu Qiao, Xizhou Zhu

In this paper, we focus on monolithic Multimodal Large Language Models (MLLMs) that integrate visual encoding and language decoding into a single LLM. In particular, we identify that existing pre-training strategies for monolithic MLLMs often suffer from unstable optimization or catastrophic forgetting. To address this issue, our core idea is to embed a new visual parameter space into a pre-trained LLM, thereby stably learning visual knowledge from noisy data while freezing the LLM. Based on this principle, we present Mono-InternVL, a novel monolithic MLLM that seamlessly integrates a set of visual experts via a multimodal mixture-of-experts structure. Moreover, we propose an innovative pre-training strategy to maximize the visual capability of Mono-InternVL, namely Endogenous Visual Pre-training (EViP). In particular, EViP is designed as a progressive learning process for visual experts, which aims to fully exploit the visual knowledge from noisy data to high-quality data. To validate our approach, we conduct extensive experiments on 16 benchmarks. Experimental results confirm the superior performance of Mono-InternVL than existing monolithic MLLMs on 13 of 16 multimodal benchmarks, e.g., +80 points over Emu3 on OCRBench. Compared to the modular baseline, i.e., InternVL-1.5, Mono-InternVL still retains comparable multimodal performance while reducing up to 67% first token latency. Code and model are released at https://github.com/OpenGVLab/Mono-InternVL.

comment: Accepted by CVPR 2025

♻ ★ Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models

Qiguang Chen, Libo Qin, Jinhao Liu, Dengyun Peng, Jiannan Guan, Peng Wang, Mengkang Hu, Yuhang Zhou, Te Gao, Wanxiang Che

Recent advancements in reasoning with large language models (RLLMs), such as OpenAI-O1 and DeepSeek-R1, have demonstrated their impressive capabilities in complex domains like mathematics and coding. A central factor in their success lies in the application of long chain-of-thought (Long CoT) characteristics, which enhance reasoning abilities and enable the solution of intricate problems. However, despite these developments, a comprehensive survey on Long CoT is still lacking, limiting our understanding of its distinctions from traditional short chain-of-thought (Short CoT) and complicating ongoing debates on issues like "overthinking" and "test-time scaling." This survey seeks to fill this gap by offering a unified perspective on Long CoT. (1) We first distinguish Long CoT from Short CoT and introduce a novel taxonomy to categorize current reasoning paradigms. (2) Next, we explore the key characteristics of Long CoT: deep reasoning, extensive exploration, and feasible reflection, which enable models to handle more complex tasks and produce more efficient, coherent outcomes compared to the shallower Short CoT. (3) We then investigate key phenomena such as the emergence of Long CoT with these characteristics, including overthinking, and test-time scaling, offering insights into how these processes manifest in practice. (4) Finally, we identify significant research gaps and highlight promising future directions, including the integration of multi-modal reasoning, efficiency improvements, and enhanced knowledge frameworks. By providing a structured overview, this survey aims to inspire future research and further the development of logical reasoning in artificial intelligence.

comment: Paper are available at https://long-cot.github.io/

♻ ★ MEDA: Dynamic KV Cache Allocation for Efficient Multimodal Long-Context Inference NAACL 2025

Zhongwei Wan, Hui Shen, Xin Wang, Che Liu, Zheda Mai, Mi Zhang

Long-context Multimodal Large Language Models (MLLMs) that incorporate long text-image and text-video modalities, demand substantial resources as their multimodal Key-Value (KV) caches grow with increasing input lengths, challenging inference efficiency. Existing methods for KV cache compression, in both text-only and multimodal LLMs, have neglected attention density variations across layers, thus often adopting uniform or progressive reduction strategies for layer-wise cache allocation. In this work, we propose MEDA, a dynamic layer-wise KV cache allocation method for efficient multimodal long-context inference. As its core, MEDA utilizes cross-modal attention entropy to determine the KV cache size at each MLLMs layer. Given the dynamically allocated KV cache size at each layer, MEDA also employs a KV pair selection scheme to identify which KV pairs to select and a KV pair merging strategy that merges the selected and non-selected ones to preserve information from the entire context. MEDA achieves up to 72% KV cache memory reduction and 2.82 times faster decoding speed, while maintaining or enhancing performance on various multimodal tasks in long-context settings, including multi-images and long-video scenarios. Our code is released at https://github.com/AIoT-MLSys-Lab/MEDA.

comment: NAACL 2025 Main

♻ ★ Multi-agent KTO: Reinforcing Strategic Interactions of Large Language Model in Language Game

Rong Ye, Yongxin Zhang, Yikai Zhang, Haoyu Kuang, Zhongyu Wei, Peng Sun

Achieving Artificial General Intelligence (AGI) requires AI agents that can not only make stratigic decisions but also engage in flexible and meaningful communication. Inspired by Wittgenstein's language game theory in Philosophical Investigations, we propose that language agents can learn through in-context interaction rather than traditional multi-stage frameworks that separate decision-making from language expression. Using Werewolf, a social deduction game that tests language understanding, strategic interaction, and adaptability, we develop the Multi-agent Kahneman & Tversky's Optimization (MaKTO). MaKTO engages diverse models in extensive gameplay to generate unpaired desirable and unacceptable responses, then employs KTO to refine the model's decision-making process. In 9-player Werewolf games, MaKTO achieves a 61% average win rate across various models, outperforming GPT-4o and two-stage RL agents by relative improvements of 23.0% and 10.9%, respectively. Notably, MaKTO also demonstrates human-like performance, winning 60% against expert players and showing only 49% detectability in Turing-style blind tests.

comment: Preprint. Code and data will be available at https://reneeye.github.io/MaKTO.html

♻ ★ Punctuation restoration improves structure understanding without supervision RepL4NLP 2025

Junghyun Min, Minho Lee, Woochul Lee, Yeonsoo Lee

Unsupervised learning objectives like autoregressive and masked language modeling constitute a significant part in producing pre-trained representations that perform various downstream applications from natural language understanding to conversational tasks. However, despite impressive generative capabilities of recent large language models, their abilities to capture syntactic or semantic structure within text lag behind. We hypothesize that the mismatch between linguistic performance and competence in machines is attributable to insufficient learning of linguistic structure knowledge via currently popular pre-training objectives. Working with English, we show that punctuation restoration as a learning objective improves performance on structure-related tasks like named entity recognition, open information extraction, chunking, and part-of-speech tagging. Punctuation restoration results in $\blacktriangle$$\geq2\%$p improvement in 16 out of 18 experiments, across 6 out of 7 tasks. Our results show that punctuation restoration is an effective learning objective that can improve structure understanding and yield a more robust structure-aware representations of natural language in base-sized models.

comment: 11 pages, 1 figure, 6 tables. RepL4NLP 2025

♻ ★ SAEBench: A Comprehensive Benchmark for Sparse Autoencoders in Language Model Interpretability

Adam Karvonen, Can Rager, Johnny Lin, Curt Tigges, Joseph Bloom, David Chanin, Yeu-Tong Lau, Eoin Farrell, Callum McDougall, Kola Ayonrinde, Matthew Wearden, Arthur Conmy, Samuel Marks, Neel Nanda

Sparse autoencoders (SAEs) are a popular technique for interpreting language model activations, and there is extensive recent work on improving SAE effectiveness. However, most prior work evaluates progress using unsupervised proxy metrics with unclear practical relevance. We introduce SAEBench, a comprehensive evaluation suite that measures SAE performance across seven diverse metrics, spanning interpretability, feature disentanglement and practical applications like unlearning. To enable systematic comparison, we open-source a suite of over 200 SAEs across eight recently proposed SAE architectures and training algorithms. Our evaluation reveals that gains on proxy metrics do not reliably translate to better practical performance. For instance, while Matryoshka SAEs slightly underperform on existing proxy metrics, they substantially outperform other architectures on feature disentanglement metrics; moreover, this advantage grows with SAE scale. By providing a standardized framework for measuring progress in SAE development, SAEBench enables researchers to study scaling trends and make nuanced comparisons between different SAE architectures and training methodologies. Our interactive interface enables researchers to flexibly visualize relationships between metrics across hundreds of open-source SAEs at: https://saebench.xyz

♻ ★ D2O: Dynamic Discriminative Operations for Efficient Long-Context Inference of Large Language Models ICLR 2025

Zhongwei Wan, Xinjian Wu, Yu Zhang, Yi Xin, Chaofan Tao, Zhihong Zhu, Xin Wang, Siqi Luo, Jing Xiong, Longyue Wang, Mi Zhang

Generative inference in Large Language Models (LLMs) is impeded by the growing memory demands of Key-Value (KV) cache, especially for longer sequences. Traditional KV cache eviction strategies, which discard less critical KV pairs based on attention scores, often degrade generation quality, leading to issues such as context loss or hallucinations. In this work, we introduce Dynamic Discriminative Operations (D2O), a KV cache compression method that optimizes KV cache size dynamically and discriminatively at two levels without fine-tuning, while preserving essential context. At layer level, D2O leverages the varying densities of attention weights between shallow and deep layers to dynamically determine which layers should avoid excessive eviction via a novel dynamic allocation strategy to minimize information loss. At token level, D2O incorporates a compensation mechanism that maintains a similarity threshold to re-discriminate the importance of currently discarded tokens, determining whether they should be recalled and merged with similar tokens. We conduct experiments on various benchmarks and LLM architectures. Our results show that D2O not only achieves significant memory savings and enhances inference throughput by more than 3$\times$ but also maintains high-quality long-text generation.

comment: ICLR 2025

♻ ★ Grounding Natural Language to SQL Translation with Data-Based Self-Explanations ICDE2025

Yuankai Fan, Tonghui Ren, Can Huang, Zhenying He, X. Sean Wang

Natural Language Interfaces for Databases empower non-technical users to interact with data using natural language (NL). Advanced approaches, utilizing either neural sequence-to-sequence or more recent sophisticated large-scale language models, typically implement NL to SQL (NL2SQL) translation in an end-to-end fashion. However, like humans, these end-to-end translation models may not always generate the best SQL output on their first try. In this paper, we propose CycleSQL, an iterative framework designed for end-to-end translation models to autonomously generate the best output through self-evaluation. The main idea of CycleSQL is to introduce data-grounded NL explanations of query results as self-provided feedback, and use the feedback to validate the correctness of the translation iteratively, hence improving the overall translation accuracy. Extensive experiments, including quantitative and qualitative evaluations, are conducted to study CycleSQL by applying it to seven existing translation models on five widely used benchmarks. The results show that 1) the feedback loop introduced in CycleSQL can consistently improve the performance of existing models, and in particular, by applying CycleSQL to RESDSQL, obtains a translation accuracy of 82.0% (+2.6%) on the validation set, and 81.6% (+3.2%) on the test set of Spider benchmark; 2) the generated NL explanations can also provide insightful information for users, aiding in the comprehension of translation results and consequently enhancing the interpretability of NL2SQL translation.

comment: ICDE2025

♻ ★ Enhancing Chain of Thought Prompting in Large Language Models via Reasoning Patterns

Yufeng Zhang, Xuepeng Wang, Lingxiang Wu, Jinqiao Wang

Chain of Thought (CoT) prompting can encourage language models to engage in multi-step logical reasoning. The quality of the provided demonstrations significantly influences the success of downstream inference tasks. Current unsupervised CoT methods primarily select examples based on the semantics of the questions, which can introduce noise and lack interpretability. In this paper, we propose leveraging reasoning patterns to enhance CoT prompting effectiveness. Reasoning patterns represent the process by which language models arrive at their final results. By utilizing prior knowledge and prompt-based methods from large models, we first construct task-specific pattern sets. We then select diverse demonstrations based on different reasoning patterns. This approach not only mitigates the impact of noise but also provides explicit interpretability to help us understand the mechanisms of CoT. Extensive experiments demonstrate that our method is more robust and consistently leads to improvements across various reasoning tasks.

♻ ★ Preference Alignment for Diffusion Model via Explicit Denoised Distribution Estimation

Dingyuan Shi, Yong Wang, Hangyu Li, Xiangxiang Chu

Diffusion models have shown remarkable success in text-to-image generation, making preference alignment for these models increasingly important. The preference labels are typically available only at the terminal of denoising trajectories, which poses challenges in optimizing the intermediate denoising steps. In this paper, we propose to conduct Denoised Distribution Estimation (DDE) that explicitly connects intermediate steps to the terminal denoised distribution. Therefore, preference labels can be used for the entire trajectory optimization. To this end, we design two estimation strategies for our DDE. The first is stepwise estimation, which utilizes the conditional denoised distribution to estimate the model denoised distribution. The second is single-shot estimation, which converts the model output into the terminal denoised distribution via DDIM modeling. Analytically and empirically, we reveal that DDE equipped with two estimation strategies naturally derives a novel credit assignment scheme that prioritizes optimizing the middle part of the denoising trajectory. Extensive experiments demonstrate that our approach achieves superior performance, both quantitatively and qualitatively.

♻ ★ MedHallBench: A New Benchmark for Assessing Hallucination in Medical Large Language Models AAAI-25

Kaiwen Zuo, Yirui Jiang

Medical Large Language Models (MLLMs) have demonstrated potential in healthcare applications, yet their propensity for hallucinations -- generating medically implausible or inaccurate information -- presents substantial risks to patient care. This paper introduces MedHallBench, a comprehensive benchmark framework for evaluating and mitigating hallucinations in MLLMs. Our methodology integrates expert-validated medical case scenarios with established medical databases to create a robust evaluation dataset. The framework employs a sophisticated measurement system that combines automated ACHMI (Automatic Caption Hallucination Measurement in Medical Imaging) scoring with rigorous clinical expert evaluations and utilizes reinforcement learning methods to achieve automatic annotation. Through an optimized reinforcement learning from human feedback (RLHF) training pipeline specifically designed for medical applications, MedHallBench enables thorough evaluation of MLLMs across diverse clinical contexts while maintaining stringent accuracy standards. We conducted comparative experiments involving various models, utilizing the benchmark to establish a baseline for widely adopted large language models (LLMs). Our findings indicate that ACHMI provides a more nuanced understanding of the effects of hallucinations compared to traditional metrics, thereby highlighting its advantages in hallucination assessment. This research establishes a foundational framework for enhancing MLLMs' reliability in healthcare settings and presents actionable strategies for addressing the critical challenge of AI hallucinations in medical applications.

comment: Published to AAAI-25 Bridge Program

★ Studying Classifier(-Free) Guidance From a Classifier-Centric Perspective

Xiaoming Zhao, Alexander G. Schwing

Classifier-free guidance has become a staple for conditional generation with denoising diffusion models. However, a comprehensive understanding of classifier-free guidance is still missing. In this work, we carry out an empirical study to provide a fresh perspective on classifier-free guidance. Concretely, instead of solely focusing on classifier-free guidance, we trace back to the root, i.e., classifier guidance, pinpoint the key assumption for the derivation, and conduct a systematic study to understand the role of the classifier. We find that both classifier guidance and classifier-free guidance achieve conditional generation by pushing the denoising diffusion trajectories away from decision boundaries, i.e., areas where conditional information is usually entangled and is hard to learn. Based on this classifier-centric understanding, we propose a generic postprocessing step built upon flow-matching to shrink the gap between the learned distribution for a pre-trained denoising diffusion model and the real data distribution, majorly around the decision boundaries. Experiments on various datasets verify the effectiveness of the proposed approach.

★ The Curse of Conditions: Analyzing and Improving Optimal Transport for Conditional Flow-Based Generation

Ho Kei Cheng, Alexander Schwing

Minibatch optimal transport coupling straightens paths in unconditional flow matching. This leads to computationally less demanding inference as fewer integration steps and less complex numerical solvers can be employed when numerically solving an ordinary differential equation at test time. However, in the conditional setting, minibatch optimal transport falls short. This is because the default optimal transport mapping disregards conditions, resulting in a conditionally skewed prior distribution during training. In contrast, at test time, we have no access to the skewed prior, and instead sample from the full, unbiased prior distribution. This gap between training and testing leads to a subpar performance. To bridge this gap, we propose conditional optimal transport C^2OT that adds a conditional weighting term in the cost matrix when computing the optimal transport assignment. Experiments demonstrate that this simple fix works with both discrete and continuous conditions in 8gaussians-to-moons, CIFAR-10, ImageNet-32x32, and ImageNet-256x256. Our method performs better overall compared to the existing baselines across different function evaluation budgets. Code is available at https://hkchengrex.github.io/C2OT

comment: Project page: https://hkchengrex.github.io/C2OT

★ A Frustratingly Simple Yet Highly Effective Attack Baseline: Over 90% Success Rate Against the Strong Black-box Models of GPT-4.5/4o/o1

Zhaoyi Li, Xiaohan Zhao, Dong-Dong Wu, Jiacheng Cui, Zhiqiang Shen

Despite promising performance on open-source large vision-language models (LVLMs), transfer-based targeted attacks often fail against black-box commercial LVLMs. Analyzing failed adversarial perturbations reveals that the learned perturbations typically originate from a uniform distribution and lack clear semantic details, resulting in unintended responses. This critical absence of semantic information leads commercial LVLMs to either ignore the perturbation entirely or misinterpret its embedded semantics, thereby causing the attack to fail. To overcome these issues, we notice that identifying core semantic objects is a key objective for models trained with various datasets and methodologies. This insight motivates our approach that refines semantic clarity by encoding explicit semantic details within local regions, thus ensuring interoperability and capturing finer-grained features, and by concentrating modifications on semantically rich areas rather than applying them uniformly. To achieve this, we propose a simple yet highly effective solution: at each optimization step, the adversarial image is cropped randomly by a controlled aspect ratio and scale, resized, and then aligned with the target image in the embedding space. Experimental results confirm our hypothesis. Our adversarial examples crafted with local-aggregated perturbations focused on crucial regions exhibit surprisingly good transferability to commercial LVLMs, including GPT-4.5, GPT-4o, Gemini-2.0-flash, Claude-3.5-sonnet, Claude-3.7-sonnet, and even reasoning models like o1, Claude-3.7-thinking and Gemini-2.0-flash-thinking. Our approach achieves success rates exceeding 90% on GPT-4.5, 4o, and o1, significantly outperforming all prior state-of-the-art attack methods. Our optimized adversarial examples under different configurations and training code are available at https://github.com/VILA-Lab/M-Attack.

comment: Code at: https://github.com/VILA-Lab/M-Attack

★ Charting and Navigating Hugging Face's Model Atlas

Eliahu Horwitz, Nitzan Kurer, Jonathan Kahana, Liel Amar, Yedid Hoshen

As there are now millions of publicly available neural networks, searching and analyzing large model repositories becomes increasingly important. Navigating so many models requires an atlas, but as most models are poorly documented charting such an atlas is challenging. To explore the hidden potential of model repositories, we chart a preliminary atlas representing the documented fraction of Hugging Face. It provides stunning visualizations of the model landscape and evolution. We demonstrate several applications of this atlas including predicting model attributes (e.g., accuracy), and analyzing trends in computer vision models. However, as the current atlas remains incomplete, we propose a method for charting undocumented regions. Specifically, we identify high-confidence structural priors based on dominant real-world model training practices. Leveraging these priors, our approach enables accurate mapping of previously undocumented areas of the atlas. We publicly release our datasets, code, and interactive atlas.

★ Kolmogorov-Arnold Attention: Is Learnable Attention Better For Vision Transformers?

Subhajit Maity, Killian Hitsman, Xin Li, Aritra Dutta

Kolmogorov-Arnold networks (KANs) are a remarkable innovation consisting of learnable activation functions with the potential to capture more complex relationships from data. Although KANs are useful in finding symbolic representations and continual learning of one-dimensional functions, their effectiveness in diverse machine learning (ML) tasks, such as vision, remains questionable. Presently, KANs are deployed by replacing multilayer perceptrons (MLPs) in deep network architectures, including advanced architectures such as vision Transformers (ViTs). In this paper, we are the first to design a general learnable Kolmogorov-Arnold Attention (KArAt) for vanilla ViTs that can operate on any choice of basis. However, the computing and memory costs of training them motivated us to propose a more modular version, and we designed particular learnable attention, called Fourier-KArAt. Fourier-KArAt and its variants either outperform their ViT counterparts or show comparable performance on CIFAR-10, CIFAR-100, and ImageNet-1K datasets. We dissect these architectures' performance and generalization capacity by analyzing their loss landscapes, weight distributions, optimizer path, attention visualization, and spectral behavior, and contrast them with vanilla ViTs. The goal of this paper is not to produce parameter- and compute-efficient attention, but to encourage the community to explore KANs in conjunction with more advanced architectures that require a careful understanding of learnable activations. Our open-source code and implementation details are available on: https://subhajitmaity.me/KArAt

comment: Preprint, Appendix included

★ Uncertainty in Action: Confidence Elicitation in Embodied Agents

Tianjiao Yu, Vedant Shah, Muntasir Wahed, Kiet A. Nguyen, Adheesh Juvekar, Tal August, Ismini Lourentzou

Expressing confidence is challenging for embodied agents navigating dynamic multimodal environments, where uncertainty arises from both perception and decision-making processes. We present the first work investigating embodied confidence elicitation in open-ended multimodal environments. We introduce Elicitation Policies, which structure confidence assessment across inductive, deductive, and abductive reasoning, along with Execution Policies, which enhance confidence calibration through scenario reinterpretation, action sampling, and hypothetical reasoning. Evaluating agents in calibration and failure prediction tasks within the Minecraft environment, we show that structured reasoning approaches, such as Chain-of-Thoughts, improve confidence calibration. However, our findings also reveal persistent challenges in distinguishing uncertainty, particularly under abductive settings, underscoring the need for more sophisticated embodied confidence elicitation methods.

comment: Project page: https://plan-lab.github.io/ece/

★ NIL: No-data Imitation Learning by Leveraging Pre-trained Video Diffusion Models

Mert Albaba, Chenhao Li, Markos Diomataris, Omid Taheri, Andreas Krause, Michael Black

Acquiring physically plausible motor skills across diverse and unconventional morphologies-including humanoid robots, quadrupeds, and animals-is essential for advancing character simulation and robotics. Traditional methods, such as reinforcement learning (RL) are task- and body-specific, require extensive reward function engineering, and do not generalize well. Imitation learning offers an alternative but relies heavily on high-quality expert demonstrations, which are difficult to obtain for non-human morphologies. Video diffusion models, on the other hand, are capable of generating realistic videos of various morphologies, from humans to ants. Leveraging this capability, we propose a data-independent approach for skill acquisition that learns 3D motor skills from 2D-generated videos, with generalization capability to unconventional and non-human forms. Specifically, we guide the imitation learning process by leveraging vision transformers for video-based comparisons by calculating pair-wise distance between video embeddings. Along with video-encoding distance, we also use a computed similarity between segmented video frames as a guidance reward. We validate our method on locomotion tasks involving unique body configurations. In humanoid robot locomotion tasks, we demonstrate that 'No-data Imitation Learning' (NIL) outperforms baselines trained on 3D motion-capture data. Our results highlight the potential of leveraging generative video models for physically plausible skill learning with diverse morphologies, effectively replacing data collection with data generation for imitation learning.

★ Transformers without Normalization CVPR 2025

Jiachen Zhu, Xinlei Chen, Kaiming He, Yann LeCun, Zhuang Liu

Normalization layers are ubiquitous in modern neural networks and have long been considered essential. This work demonstrates that Transformers without normalization can achieve the same or better performance using a remarkably simple technique. We introduce Dynamic Tanh (DyT), an element-wise operation $DyT($x$) = \tanh(\alpha $x$)$, as a drop-in replacement for normalization layers in Transformers. DyT is inspired by the observation that layer normalization in Transformers often produces tanh-like, $S$-shaped input-output mappings. By incorporating DyT, Transformers without normalization can match or exceed the performance of their normalized counterparts, mostly without hyperparameter tuning. We validate the effectiveness of Transformers with DyT across diverse settings, ranging from recognition to generation, supervised to self-supervised learning, and computer vision to language models. These findings challenge the conventional understanding that normalization layers are indispensable in modern neural networks, and offer new insights into their role in deep networks.

comment: CVPR 2025; Project page: https://jiachenzhu.github.io/DyT/

★ Poly-MgNet: Polynomial Building Blocks in Multigrid-Inspired ResNets

Antonia van Betteray, Matthias Rottmann, Karsten Kahl

The structural analogies of ResNets and Multigrid (MG) methods such as common building blocks like convolutions and poolings where already pointed out by He et al.\ in 2016. Multigrid methods are used in the context of scientific computing for solving large sparse linear systems arising from partial differential equations. MG methods particularly rely on two main concepts: smoothing and residual restriction / coarsening. Exploiting these analogies, He and Xu developed the MgNet framework, which integrates MG schemes into the design of ResNets. In this work, we introduce a novel neural network building block inspired by polynomial smoothers from MG theory. Our polynomial block from an MG perspective naturally extends the MgNet framework to Poly-Mgnet and at the same time reduces the number of weights in MgNet. We present a comprehensive study of our polynomial block, analyzing the choice of initial coefficients, the polynomial degree, the placement of activation functions, as well as of batch normalizations. Our results demonstrate that constructing (quadratic) polynomial building blocks based on real and imaginary polynomial roots enhances Poly-MgNet's capacity in terms of accuracy. Furthermore, our approach achieves an improved trade-off of model accuracy and number of weights compared to ResNet as well as compared to specific configurations of MgNet.

★ The Spectral Bias of Shallow Neural Network Learning is Shaped by the Choice of Non-linearity

Justin Sahs, Ryan Pyle, Fabio Anselmi, Ankit Patel

Despite classical statistical theory predicting severe overfitting, modern massively overparameterized neural networks still generalize well. This unexpected property is attributed to the network's so-called implicit bias, which describes its propensity to converge to solutions that generalize effectively, among the many possible that correctly label the training data. The aim of our research is to explore this bias from a new perspective, focusing on how non-linear activation functions contribute to shaping it. First, we introduce a reparameterization which removes a continuous weight rescaling symmetry. Second, in the kernel regime, we leverage this reparameterization to generalize recent findings that relate shallow Neural Networks to the Radon transform, deriving an explicit formula for the implicit bias induced by a broad class of activation functions. Specifically, by utilizing the connection between the Radon transform and the Fourier transform, we interpret the kernel regime's inductive bias as minimizing a spectral seminorm that penalizes high-frequency components, in a manner dependent on the activation function. Finally, in the adaptive regime, we demonstrate the existence of local dynamical attractors that facilitate the formation of clusters of hyperplanes where the input to a neuron's activation function is zero, yielding alignment between many neurons' response functions. We confirm these theoretical results with simulations. All together, our work provides a deeper understanding of the mechanisms underlying the generalization capabilities of overparameterized neural networks and its relation with the implicit bias, offering potential pathways for designing more efficient and robust models.

comment: 18 pages, 10 figures in main text

★ On the Injective Norm of Sums of Random Tensors and the Moments of Gaussian Chaoses

Ishaq Aden-Ali

We prove an upper bound on the expected $\ell_p$ injective norm of sums of subgaussian random tensors. Our proof is simple and does not rely on any explicit geometric or chaining arguments. Instead, it follows from a simple application of the PAC-Bayesian lemma, a tool that has proven effective at controlling the suprema of certain ``smooth'' empirical processes in recent years. Our bound strictly improves a very recent result of Bandeira, Gopi, Jiang, Lucca, and Rothvoss. In the Euclidean case ($p=2$), our bound sharpens a result of Lata{\l}a that was central to proving his estimates on the moments of Gaussian chaoses. As a consequence, we obtain an elementary proof of this fundamental result.

comment: 12 pages

★ Sample and Map from a Single Convex Potential: Generation using Conjugate Moment Measures

Nina Vesseron, Louis Béthune, Marco Cuturi

A common approach to generative modeling is to split model-fitting into two blocks: define first how to sample noise (e.g. Gaussian) and choose next what to do with it (e.g. using a single map or flows). We explore in this work an alternative route that ties sampling and mapping. We find inspiration in moment measures, a result that states that for any measure $\rho$ supported on a compact convex set of $\mathbb{R}^d$, there exists a unique convex potential $u$ such that $\rho=\nabla u\,\sharp\,e^{-u}$. While this does seem to tie effectively sampling (from log-concave distribution $e^{-u}$) and action (pushing particles through $\nabla u$), we observe on simple examples (e.g., Gaussians or 1D distributions) that this choice is ill-suited for practical tasks. We study an alternative factorization, where $\rho$ is factorized as $\nabla w^*\,\sharp\,e^{-w}$, where $w^*$ is the convex conjugate of $w$. We call this approach conjugate moment measures, and show far more intuitive results on these examples. Because $\nabla w^*$ is the Monge map between the log-concave distribution $e^{-w}$ and $\rho$, we rely on optimal transport solvers to propose an algorithm to recover $w$ from samples of $\rho$, and parameterize $w$ as an input-convex neural network.

★ Unveiling the Mathematical Reasoning in DeepSeek Models: A Comparative Study of Large Language Models

Afrar Jahin, Arif Hassan Zidan, Yu Bao, Shizhe Liang, Tianming Liu, Wei Zhang

With the rapid evolution of Artificial Intelligence (AI), Large Language Models (LLMs) have reshaped the frontiers of various fields, spanning healthcare, public health, engineering, science, agriculture, education, arts, humanities, and mathematical reasoning. Among these advancements, DeepSeek models have emerged as noteworthy contenders, demonstrating promising capabilities that set them apart from their peers. While previous studies have conducted comparative analyses of LLMs, few have delivered a comprehensive evaluation of mathematical reasoning across a broad spectrum of LLMs. In this work, we aim to bridge this gap by conducting an in-depth comparative study, focusing on the strengths and limitations of DeepSeek models in relation to their leading counterparts. In particular, our study systematically evaluates the mathematical reasoning performance of two DeepSeek models alongside five prominent LLMs across three independent benchmark datasets. The findings reveal several key insights: 1). DeepSeek-R1 consistently achieved the highest accuracy on two of the three datasets, demonstrating strong mathematical reasoning capabilities. 2). The distilled variant of LLMs significantly underperformed compared to its peers, highlighting potential drawbacks in using distillation techniques. 3). In terms of response time, Gemini 2.0 Flash demonstrated the fastest processing speed, outperforming other models in efficiency, which is a crucial factor for real-time applications. Beyond these quantitative assessments, we delve into how architecture, training, and optimization impact LLMs' mathematical reasoning. Moreover, our study goes beyond mere performance comparison by identifying key areas for future advancements in LLM-driven mathematical reasoning. This research enhances our understanding of LLMs' mathematical reasoning and lays the groundwork for future advancements

★ Radar: Fast Long-Context Decoding for Any Transformer ICLR 2025

Yongchang Hao, Mengyao Zhai, Hossein Hajimirsadeghi, Sepidehsadat Hosseini, Frederick Tung

Transformer models have demonstrated exceptional performance across a wide range of applications. Though forming the foundation of Transformer models, the dot-product attention does not scale well to long-context data since its time requirement grows quadratically with context length. In this work, we propose Radar, a training-free approach that accelerates inference by dynamically searching for the most important context tokens. For any pre-trained Transformer, Radar can reduce the decoding time complexity without training or heuristically evicting tokens. Moreover, we provide theoretical justification for our approach, demonstrating that Radar can reliably identify the most important tokens with high probability. We conduct extensive comparisons with the previous methods on a wide range of tasks. The results demonstrate that Radar achieves the state-of-the-art performance across different architectures with reduced time complexity, offering a practical solution for efficient long-context processing of Transformers.

comment: Accepted @ ICLR 2025

★ FedPCA: Noise-Robust Fair Federated Learning via Performance-Capacity Analysis

Nannan Wu, Zengqiang Yan, Nong Sang, Li Yu, Chang Wen Chen

Training a model that effectively handles both common and rare data-i.e., achieving performance fairness-is crucial in federated learning (FL). While existing fair FL methods have shown effectiveness, they remain vulnerable to mislabeled data. Ensuring robustness in fair FL is therefore essential. However, fairness and robustness inherently compete, which causes robust strategies to hinder fairness. In this paper, we attribute this competition to the homogeneity in loss patterns exhibited by rare and mislabeled data clients, preventing existing loss-based fair and robust FL methods from effectively distinguishing and handling these two distinct client types. To address this, we propose performance-capacity analysis, which jointly considers model performance on each client and its capacity to handle the dataset, measured by loss and a newly introduced feature dispersion score. This allows mislabeled clients to be identified by their significantly deviated performance relative to capacity while preserving rare data clients. Building on this, we introduce FedPCA, an FL method that robustly achieves fairness. FedPCA first identifies mislabeled clients via a Gaussian Mixture Model on loss-dispersion pairs, then applies fairness and robustness strategies in global aggregation and local training by adjusting client weights and selectively using reliable data. Extensive experiments on three datasets demonstrate FedPCA's effectiveness in tackling this complex challenge. Code will be publicly available upon acceptance.

comment: Preprint

★ ASIDE: Architectural Separation of Instructions and Data in Language Models ICLR 2025

Egor Zverev, Evgenii Kortukov, Alexander Panfilov, Soroush Tabesh, Alexandra Volkova, Sebastian Lapuschkin, Wojciech Samek, Christoph H. Lampert

Despite their remarkable performance, large language models lack elementary safety features, and this makes them susceptible to numerous malicious attacks. In particular, previous work has identified the absence of an intrinsic separation between instructions and data as a root cause for the success of prompt injection attacks. In this work, we propose an architectural change, ASIDE, that allows the model to clearly separate between instructions and data by using separate embeddings for them. Instead of training the embeddings from scratch, we propose a method to convert an existing model to ASIDE form by using two copies of the original model's embeddings layer, and applying an orthogonal rotation to one of them. We demonstrate the effectiveness of our method by showing (1) highly increased instruction-data separation scores without a loss in model capabilities and (2) competitive results on prompt injection benchmarks, even without dedicated safety training. Additionally, we study the working mechanism behind our method through an analysis of model representations.

comment: ICLR 2025 Workshop on Building Trust in Language Models and Applications

★ From Linear to Spline-Based Classification:Developing and Enhancing SMPA for Noisy Non-Linear Datasets

Vatsal Srivastava

Building upon the concepts and mechanisms used for the development in Moving Points Algorithm, we will now explore how non linear decision boundaries can be developed for classification tasks. First we will look at the classification performance of MPA and some minor developments in the original algorithm. We then discuss the concepts behind using cubic splines for classification with a similar learning mechanism and finally analyze training results on synthetic datasets with known properties.

★ DP-GPL: Differentially Private Graph Prompt Learning

Jing Xu, Franziska Boenisch, Iyiola Emmanuel Olatunji, Adam Dziedzic

Graph Neural Networks (GNNs) have shown remarkable performance in various applications. Recently, graph prompt learning has emerged as a powerful GNN training paradigm, inspired by advances in language and vision foundation models. Here, a GNN is pre-trained on public data and then adapted to sensitive tasks using lightweight graph prompts. However, using prompts from sensitive data poses privacy risks. In this work, we are the first to investigate these practical risks in graph prompts by instantiating a membership inference attack that reveals significant privacy leakage. We also find that the standard privacy method, DP-SGD, fails to provide practical privacy-utility trade-offs in graph prompt learning, likely due to the small number of sensitive data points used to learn the prompts. As a solution, we propose DP-GPL for differentially private graph prompt learning based on the PATE framework, that generates a graph prompt with differential privacy guarantees. Our evaluation across various graph prompt learning methods, GNN architectures, and pre-training strategies demonstrates that our algorithm achieves high utility at strong privacy, effectively mitigating privacy concerns while preserving the powerful capabilities of prompted GNNs as powerful foundation models in the graph domain.

★ Language Models, Graph Searching, and Supervision Adulteration: When More Supervision is Less and How to Make More More ICLR 2025

Arvid Frydenlund

This work concerns the path-star task, a minimal example of searching over a graph. The graph, $G$, is star-shaped with $D$ arms radiating from a start node, $s$. A language model (LM) is given $G$, $s$, and a target node $t$, which ends one of the arms and is tasked with generating the arm containing $t$. The minimal nature of this task means only a single choice needs to be made: which of the $D$ arms contains $t$? Decoder-only LMs fail to solve this elementary task above $1/D$ chance due to a learned shortcut that absorbs training supervision. We show how this pathology is caused by excess supervision and we present a series of solutions demonstrating that the task is solvable via decoder-only LMs. We find that the task's minimal nature causes its difficulty, as it prevents task decomposition. Our solutions provide insight into the pathology and its implications for LMs trained via next-token prediction.

comment: A reduced version of this work has been accepted to the Workshop on Spurious Correlation and Shortcut Learning: Foundations and Solutions (SCSL) at ICLR 2025. Full version under review

★ GBSVR: Granular Ball Support Vector Regression

Reshma Rastogi, Ankush Bisht, Sanjay Kumar, Suresh Chandra

Support Vector Regression (SVR) and its variants are widely used to handle regression tasks, however, since their solution involves solving an expensive quadratic programming problem, it limits its application, especially when dealing with large datasets. Additionally, SVR uses an epsilon-insensitive loss function which is sensitive to outliers and therefore can adversely affect its performance. We propose Granular Ball Support Vector Regression (GBSVR) to tackle problem of regression by using granular ball concept. These balls are useful in simplifying complex data spaces for machine learning tasks, however, to the best of our knowledge, they have not been sufficiently explored for regression problems. Granular balls group the data points into balls based on their proximity and reduce the computational cost in SVR by replacing the large number of data points with far fewer granular balls. This work also suggests a discretization method for continuous-valued attributes to facilitate the construction of granular balls. The effectiveness of the proposed approach is evaluated on several benchmark datasets and it outperforms existing state-of-the-art approaches

★ Structured Preconditioners in Adaptive Optimization: A Unified Analysis

Shuo Xie, Tianhao Wang, Sashank Reddi, Sanjiv Kumar, Zhiyuan Li

We present a novel unified analysis for a broad class of adaptive optimization algorithms with structured (e.g., layerwise, diagonal, and kronecker-factored) preconditioners for both online regret minimization and offline convex optimization. Our analysis not only provides matching rate to several important structured preconditioned algorithms including diagonal AdaGrad, full-matrix AdaGrad, and AdaGrad-Norm, but also gives an improved convergence rate for a one-sided variant of Shampoo over that of original Shampoo. Interestingly, more structured preconditioners (e.g., diagonal Adagrad, AdaGrad-Norm which use less space and compute) are often presented as computationally efficient approximations to full-matrix Adagrad, aiming for improved optimization performance through better approximations. Our unified analysis challenges this prevailing view and reveals, perhaps surprisingly, that more structured preconditioners, despite using less space and computation per step, can outperform their less structured counterparts. To demonstrate this, we show that one-sided Shampoo, which is relatively much cheaper than full-matrix AdaGrad could outperform it both theoretically and experimentally.

★ AudioX: Diffusion Transformer for Anything-to-Audio Generation

Zeyue Tian, Yizhu Jin, Zhaoyang Liu, Ruibin Yuan, Xu Tan, Qifeng Chen, Wei Xue, Yike Guo

Audio and music generation have emerged as crucial tasks in many applications, yet existing approaches face significant limitations: they operate in isolation without unified capabilities across modalities, suffer from scarce high-quality, multi-modal training data, and struggle to effectively integrate diverse inputs. In this work, we propose AudioX, a unified Diffusion Transformer model for Anything-to-Audio and Music Generation. Unlike previous domain-specific models, AudioX can generate both general audio and music with high quality, while offering flexible natural language control and seamless processing of various modalities including text, video, image, music, and audio. Its key innovation is a multi-modal masked training strategy that masks inputs across modalities and forces the model to learn from masked inputs, yielding robust and unified cross-modal representations. To address data scarcity, we curate two comprehensive datasets: vggsound-caps with 190K audio captions based on the VGGSound dataset, and V2M-caps with 6 million music captions derived from the V2M dataset. Extensive experiments demonstrate that AudioX not only matches or outperforms state-of-the-art specialized models, but also offers remarkable versatility in handling diverse input modalities and generation tasks within a unified architecture. The code and datasets will be available at https://zeyuet.github.io/AudioX/

comment: The code and datasets will be available at https://zeyuet.github.io/AudioX/

★ CountPath: Automating Fragment Counting in Digital Pathology

Ana Beatriz Vieira, Maria Valente, Diana Montezuma, Tomé Albuquerque, Liliana Ribeiro, Domingos Oliveira, João Monteiro, Sofia Gonçalves, Isabel M. Pinto, Jaime S. Cardoso, Arlindo L. Oliveira

Quality control of medical images is a critical component of digital pathology, ensuring that diagnostic images meet required standards. A pre-analytical task within this process is the verification of the number of specimen fragments, a process that ensures that the number of fragments on a slide matches the number documented in the macroscopic report. This step is important to ensure that the slides contain the appropriate diagnostic material from the grossing process, thereby guaranteeing the accuracy of subsequent microscopic examination and diagnosis. Traditionally, this assessment is performed manually, requiring significant time and effort while being subject to significant variability due to its subjective nature. To address these challenges, this study explores an automated approach to fragment counting using the YOLOv9 and Vision Transformer models. Our results demonstrate that the automated system achieves a level of performance comparable to expert assessments, offering a reliable and efficient alternative to manual counting. Additionally, we present findings on interobserver variability, showing that the automated approach achieves an accuracy of 86%, which falls within the range of variation observed among experts (82-88%), further supporting its potential for integration into routine pathology workflows.

comment: 10 pages, 3 figures

★ Conformal Prediction Sets for Deep Generative Models via Reduction to Conformal Regression

Hooman Shahrokhi, Devjeet Raj Roy, Yan Yan, Venera Arnaoudova, Janaradhan Rao Doppa

We consider the problem of generating valid and small prediction sets by sampling outputs (e.g., software code and natural language text) from a black-box deep generative model for a given input (e.g., textual prompt). The validity of a prediction set is determined by a user-defined binary admissibility function depending on the target application. For example, requiring at least one program in the set to pass all test cases in code generation application. To address this problem, we develop a simple and effective conformal inference algorithm referred to as Generative Prediction Sets (GPS). Given a set of calibration examples and black-box access to a deep generative model, GPS can generate prediction sets with provable guarantees. The key insight behind GPS is to exploit the inherent structure within the distribution over the minimum number of samples needed to obtain an admissible output to develop a simple conformal regression approach over the minimum number of samples. Experiments on multiple datasets for code and math word problems using different large language models demonstrate the efficacy of GPS over state-of-the-art methods.

★ Extreme Learning Machines for Attention-based Multiple Instance Learning in Whole-Slide Image Classification

Rajiv Krishnakumar, Julien Baglio, Frederik F. Flöther, Christian Ruiz, Stefan Habringer, Nicole H. Romano

Whole-slide image classification represents a key challenge in computational pathology and medicine. Attention-based multiple instance learning (MIL) has emerged as an effective approach for this problem. However, the effect of attention mechanism architecture on model performance is not well-documented for biomedical imagery. In this work, we compare different methods and implementations of MIL, including deep learning variants. We introduce a new method using higher-dimensional feature spaces for deep MIL. We also develop a novel algorithm for whole-slide image classification where extreme machine learning is combined with attention-based MIL to improve sensitivity and reduce training complexity. We apply our algorithms to the problem of detecting circulating rare cells (CRCs), such as erythroblasts, in peripheral blood. Our results indicate that nonlinearities play a key role in the classification, as removing them leads to a sharp decrease in stability in addition to a decrease in average area under the curve (AUC) of over 4%. We also demonstrate a considerable increase in robustness of the model with improvements of over 10% in average AUC when higher-dimensional feature spaces are leveraged. In addition, we show that extreme learning machines can offer clear improvements in terms of training efficiency by reducing the number of trained parameters by a factor of 5 whilst still maintaining the average AUC to within 1.5% of the deep MIL model. Finally, we discuss options of enriching the classical computing framework with quantum algorithms in the future. This work can thus help pave the way towards more accurate and efficient single-cell diagnostics, one of the building blocks of precision medicine.

★ SySLLM: Generating Synthesized Policy Summaries for Reinforcement Learning Agents Using Large Language Models

Sahar Admoni, Omer Ben-Porat, Ofra Amir

Policies generated by Reinforcement Learning (RL) algorithms can be difficult to describe to users, as they result from the interplay between complex reward structures and neural network-based representations. This combination often leads to unpredictable behaviors, making policies challenging to analyze and posing significant obstacles to fostering human trust in real-world applications. Global policy summarization methods aim to describe agent behavior through a demonstration of actions in a subset of world-states. However, users can only watch a limited number of demonstrations, restricting their understanding of policies. Moreover, those methods overly rely on user interpretation, as they do not synthesize observations into coherent patterns. In this work, we present SySLLM (Synthesized Summary using LLMs), a novel method that employs synthesis summarization, utilizing large language models' (LLMs) extensive world knowledge and ability to capture patterns, to generate textual summaries of policies. Specifically, an expert evaluation demonstrates that the proposed approach generates summaries that capture the main insights generated by experts while not resulting in significant hallucinations. Additionally, a user study shows that SySLLM summaries are preferred over demonstration-based policy summaries and match or surpass their performance in objective agent identification tasks.

★ Sample Compression for Continual Learning

Jacob Comeau, Mathieu Bazinet, Pascal Germain, Cem Subakan

Continual learning algorithms aim to learn from a sequence of tasks, making the training distribution non-stationary. The majority of existing continual learning approaches in the literature rely on heuristics and do not provide learning guarantees for the continual learning setup. In this paper, we present a new method called 'Continual Pick-to-Learn' (CoP2L), which is able to retain the most representative samples for each task in an efficient way. The algorithm is adapted from the Pick-to-Learn algorithm, rooted in the sample compression theory. This allows us to provide high-confidence upper bounds on the generalization loss of the learned predictors, numerically computable after every update of the learned model. We also empirically show on several standard continual learning benchmarks that our algorithm is able to outperform standard experience replay, significantly mitigating catastrophic forgetting.

★ Explainable Bayesian deep learning through input-skip Latent Binary Bayesian Neural Networks

Eirik Høyheim, Lars Skaaret-Lund, Solve Sæbø, Aliaksandr Hubin

Modeling natural phenomena with artificial neural networks (ANNs) often provides highly accurate predictions. However, ANNs often suffer from over-parameterization, complicating interpretation and raising uncertainty issues. Bayesian neural networks (BNNs) address the latter by representing weights as probability distributions, allowing for predictive uncertainty evaluation. Latent binary Bayesian neural networks (LBBNNs) further handle structural uncertainty and sparsify models by removing redundant weights. This article advances LBBNNs by enabling covariates to skip to any succeeding layer or be excluded, simplifying networks and clarifying input impacts on predictions. Ultimately, a linear model or even a constant can be found to be optimal for a specific problem at hand. Furthermore, the input-skip LBBNN approach reduces network density significantly compared to standard LBBNNs, achieving over 99% reduction for small networks and over 99.9% for larger ones, while still maintaining high predictive accuracy and uncertainty measurement. For example, on MNIST, we reached 97% accuracy and great calibration with just 935 weights, reaching state-of-the-art for compression of neural networks. Furthermore, the proposed method accurately identifies the true covariates and adjusts for system non-linearity. The main contribution is the introduction of active paths, enhancing directly designed global and local explanations within the LBBNN framework, that have theoretical guarantees and do not require post hoc external tools for explanations.

comment: 44 pages, 19 tables, 25 figures. Code available at https://github.com/eirihoyh/ISLaB-LBBNN

★ Meta-learning characteristics and dynamics of quantum systems

Lucas Schorling, Pranav Vaidhyanathan, Jonas Schuff, Miguel J. Carballido, Dominik Zumbühl, Gerard Milburn, Florian Marquardt, Jakob Foerster, Michael A. Osborne, Natalia Ares

While machine learning holds great promise for quantum technologies, most current methods focus on predicting or controlling a specific quantum system. Meta-learning approaches, however, can adapt to new systems for which little data is available, by leveraging knowledge obtained from previous data associated with similar systems. In this paper, we meta-learn dynamics and characteristics of closed and open two-level systems, as well as the Heisenberg model. Based on experimental data of a Loss-DiVincenzo spin-qubit hosted in a Ge/Si core/shell nanowire for different gate voltage configurations, we predict qubit characteristics i.e. $g$-factor and Rabi frequency using meta-learning. The algorithm we introduce improves upon previous state-of-the-art meta-learning methods for physics-based systems by introducing novel techniques such as adaptive learning rates and a global optimizer for improved robustness and increased computational efficiency. We benchmark our method against other meta-learning methods, a vanilla transformer, and a multilayer perceptron, and demonstrate improved performance.

comment: 6+1 pages, 4 figures. L. Schorling and P. Vaidhyanathan contributed equally to this work

★ Representation Learning, Large-Scale 3D Molecular Pretraining, Molecular Property

Shuqi Lu, Xiaohong Ji, Bohang Zhang, Lin Yao, Siyuan Liu, Zhifeng Gao, Linfeng Zhang, Guolin Ke

Molecular pretrained representations (MPR) has emerged as a powerful approach for addressing the challenge of limited supervised data in applications such as drug discovery and material design. While early MPR methods relied on 1D sequences and 2D graphs, recent advancements have incorporated 3D conformational information to capture rich atomic interactions. However, these prior models treat molecules merely as discrete atom sets, overlooking the space surrounding them. We argue from a physical perspective that only modeling these discrete points is insufficient. We first present a simple yet insightful observation: naively adding randomly sampled virtual points beyond atoms can surprisingly enhance MPR performance. In light of this, we propose a principled framework that incorporates the entire 3D space spanned by molecules. We implement the framework via a novel Transformer-based architecture, dubbed SpaceFormer, with three key components: (1) grid-based space discretization; (2) grid sampling/merging; and (3) efficient 3D positional encoding. Extensive experiments show that SpaceFormer significantly outperforms previous 3D MPR models across various downstream tasks with limited data, validating the benefit of leveraging the additional 3D space beyond atoms in MPR models.

★ Streaming Generation of Co-Speech Gestures via Accelerated Rolling Diffusion

Evgeniia Vu, Andrei Boiarov, Dmitry Vetrov

Generating co-speech gestures in real time requires both temporal coherence and efficient sampling. We introduce Accelerated Rolling Diffusion, a novel framework for streaming gesture generation that extends rolling diffusion models with structured progressive noise scheduling, enabling seamless long-sequence motion synthesis while preserving realism and diversity. We further propose Rolling Diffusion Ladder Acceleration (RDLA), a new approach that restructures the noise schedule into a stepwise ladder, allowing multiple frames to be denoised simultaneously. This significantly improves sampling efficiency while maintaining motion consistency, achieving up to a 2x speedup with high visual fidelity and temporal coherence. We evaluate our approach on ZEGGS and BEAT, strong benchmarks for real-world applicability. Our framework is universally applicable to any diffusion-based gesture generation model, transforming it into a streaming approach. Applied to three state-of-the-art methods, it consistently outperforms them, demonstrating its effectiveness as a generalizable and efficient solution for real-time, high-fidelity co-speech gesture synthesis.

★ Applying Tabular Deep Learning Models to Estimate Crash Injury Types of Young Motorcyclists

Shriyank Somvanshi, Anannya Ghosh Tusti, Rohit Chakraborty, Subasish Das

Young motorcyclists, particularly those aged 15 to 24 years old, face a heightened risk of severe crashes due to factors such as speeding, traffic violations, and helmet usage. This study aims to identify key factors influencing crash severity by analyzing 10,726 young motorcyclist crashes in Texas from 2017 to 2022. Two advanced tabular deep learning models, ARMNet and MambaNet, were employed, using an advanced resampling technique to address class imbalance. The models were trained to classify crashes into three severity levels, Fatal or Severe, Moderate or Minor, and No Injury. ARMNet achieved an accuracy of 87 percent, outperforming 86 percent of Mambanet, with both models excelling in predicting severe and no injury crashes while facing challenges in moderate crash classification. Key findings highlight the significant influence of demographic, environmental, and behavioral factors on crash outcomes. The study underscores the need for targeted interventions, including stricter helmet enforcement and educational programs customized to young motorcyclists. These insights provide valuable guidance for policymakers in developing evidence-based strategies to enhance motorcyclist safety and reduce crash severity.

comment: 6 pages, 6 figures, accepted at IEEE CAI 2025

★ Deep Learning based discovery of Integrable Systems

Shailesh Lal, Suvajit Majumder, Evgeny Sobko

We introduce a novel machine learning based framework for discovering integrable models. Our approach first employs a synchronized ensemble of neural networks to find high-precision numerical solution to the Yang-Baxter equation within a specified class. Then, using an auxiliary system of algebraic equations, [Q_2, Q_3] = 0, and the numerical value of the Hamiltonian obtained via deep learning as a seed, we reconstruct the entire Hamiltonian family, forming an algebraic variety. We illustrate our presentation with three- and four-dimensional spin chains of difference form with local interactions. Remarkably, all discovered Hamiltonian families form rational varieties.

comment: 11 pages, 2 column text, 3 figures, Mathematica notebook with example Hamiltonians

★ OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary

Yifeng Yang, Lin Zhu, Zewen Sun, Hengyu Liu, Qinying Gu, Nanyang Ye

Out-of-distribution (OOD) detection remains challenging for deep learning models, particularly when test-time OOD samples differ significantly from training outliers. We propose OODD, a novel test-time OOD detection method that dynamically maintains and updates an OOD dictionary without fine-tuning. Our approach leverages a priority queue-based dictionary that accumulates representative OOD features during testing, combined with an informative inlier sampling strategy for in-distribution (ID) samples. To ensure stable performance during early testing, we propose a dual OOD stabilization mechanism that leverages strategically generated outliers derived from ID data. To our best knowledge, extensive experiments on the OpenOOD benchmark demonstrate that OODD significantly outperforms existing methods, achieving a 26.0% improvement in FPR95 on CIFAR-100 Far OOD detection compared to the state-of-the-art approach. Furthermore, we present an optimized variant of the KNN-based OOD detection framework that achieves a 3x speedup while maintaining detection performance.

★ SortingEnv: An Extendable RL-Environment for an Industrial Sorting Process

Tom Maus, Nico Zengeler, Tobias Glasmachers

We present a novel reinforcement learning (RL) environment designed to both optimize industrial sorting systems and study agent behavior in evolving spaces. In simulating material flow within a sorting process our environment follows the idea of a digital twin, with operational parameters like belt speed and occupancy level. To reflect real-world challenges, we integrate common upgrades to industrial setups, like new sensors or advanced machinery. It thus includes two variants: a basic version focusing on discrete belt speed adjustments and an advanced version introducing multiple sorting modes and enhanced material composition observations. We detail the observation spaces, state update mechanisms, and reward functions for both environments. We further evaluate the efficiency of common RL algorithms like Proximal Policy Optimization (PPO), Deep-Q-Networks (DQN), and Advantage Actor Critic (A2C) in comparison to a classical rule-based agent (RBA). This framework not only aids in optimizing industrial processes but also provides a foundation for studying agent behavior and transferability in evolving environments, offering insights into model performance and practical implications for real-world RL applications.

comment: Presented at the 12th International Conference on Industrial Engineering and Applications (ICIEA-EU), Munich, 2025. This article has been submitted to AIP Conference Proceedings. After it is published, it will be available in the AIP Digital Library

★ Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyond

Liang Wen, Yunke Cai, Fenrui Xiao, Xin He, Qi An, Zhenyu Duan, Yimin Du, Junchen Liu, Lifu Tang, Xiaowei Lv, Haosheng Zou, Yongchao Deng, Shousheng Jia, Xiangzheng Zhang

This paper presents our work on the Light-R1 series, with models, data, and code all released. We first focus on training long COT models from scratch, specifically starting from models initially lacking long COT capabilities. Using a curriculum training recipe consisting of two-stage SFT and semi-on-policy DPO, we train our model Light-R1-32B from Qwen2.5-32B-Instruct, resulting in superior math performance compared to DeepSeek-R1-Distill-Qwen-32B. Despite being trained exclusively on math data, Light-R1-32B shows strong generalization across other domains. In the subsequent phase of this work, we highlight the significant benefit of the 3k dataset constructed for the second SFT stage on enhancing other models. By fine-tuning DeepSeek-R1-Distilled models using this dataset, we obtain new SOTA models in 7B and 14B, while the 32B model, Light-R1-32B-DS performed comparably to QwQ-32B and DeepSeek-R1. Furthermore, we extend our work by applying reinforcement learning, specifically GRPO, on long-COT models to further improve reasoning performance. We successfully train our final Light-R1-14B-DS with RL, achieving SOTA performance among 14B parameter models in math. With AIME24 & 25 scores of 74.0 and 60.2 respectively, Light-R1-14B-DS surpasses even many 32B models and DeepSeek-R1-Distill-Llama-70B. Its RL training also exhibits well expected behavior, showing simultaneous increase in response length and reward score. The Light-R1 series of work validates training long-COT models from scratch, showcases the art in SFT data and releases SOTA models from RL.

comment: all release at https://github.com/Qihoo360/Light-R1

★ Sentiment Analysis in SemEval: A Review of Sentiment Identification Approaches

Bousselham El Haddaoui, Raddouane Chiheb, Rdouan Faizi, Abdellatif El Afia

Social media platforms are becoming the foundations of social interactions including messaging and opinion expression. In this regard, Sentiment Analysis techniques focus on providing solutions to ensure the retrieval and analysis of generated data including sentiments, emotions, and discussed topics. International competitions such as the International Workshop on Semantic Evaluation (SemEval) have attracted many researchers and practitioners with a special research interest in building sentiment analysis systems. In our work, we study top-ranking systems for each SemEval edition during the 2013-2021 period, a total of 658 teams participated in these editions with increasing interest over years. We analyze the proposed systems marking the evolution of research trends with a focus on the main components of sentiment analysis systems including data acquisition, preprocessing, and classification. Our study shows an active use of preprocessing techniques, an evolution of features engineering and word representation from lexicon-based approaches to word embeddings, and the dominance of neural networks and transformers over the classification phase fostering the use of ready-to-use models. Moreover, we provide researchers with insights based on experimented systems which will allow rapid prototyping of new systems and help practitioners build for future SemEval editions.

★ Learning Disease State from Noisy Ordinal Disease Progression Labels

Gustav Schmidt, Holger Heidrich, Philipp Berens, Sarah Müller

Learning from noisy ordinal labels is a key challenge in medical imaging. In this work, we ask whether ordinal disease progression labels (better, worse, or stable) can be used to learn a representation allowing to classify disease state. For neovascular age-related macular degeneration (nAMD), we cast the problem of modeling disease progression between medical visits as a classification task with ordinal ranks. To enhance generalization, we tailor our model to the problem setting by (1) independent image encoding, (2) antisymmetric logit space equivariance, and (3) ordinal scale awareness. In addition, we address label noise by learning an uncertainty estimate for loss re-weighting. Our approach learns an interpretable disease representation enabling strong few-shot performance for the related task of nAMD activity classification from single images, despite being trained only on image pairs with ordinal disease progression labels.

★ Finetuning Generative Trajectory Model with Reinforcement Learning from Human Feedback

Derun Li, Jianwei Ren, Yue Wang, Xin Wen, Pengxiang Li, Leimeng Xu, Kun Zhan, Zhongpu Xia, Peng Jia, Xianpeng Lang, Ningyi Xu, Hang Zhao

Generating human-like and adaptive trajectories is essential for autonomous driving in dynamic environments. While generative models have shown promise in synthesizing feasible trajectories, they often fail to capture the nuanced variability of human driving styles due to dataset biases and distributional shifts. To address this, we introduce TrajHF, a human feedback-driven finetuning framework for generative trajectory models, designed to align motion planning with diverse driving preferences. TrajHF incorporates multi-conditional denoiser and reinforcement learning with human feedback to refine multi-modal trajectory generation beyond conventional imitation learning. This enables better alignment with human driving preferences while maintaining safety and feasibility constraints. TrajHF achieves PDMS of 93.95 on NavSim benchmark, significantly exceeding other methods. TrajHF sets a new paradigm for personalized and adaptable trajectory generation in autonomous driving.

comment: 10 pages, 5 figures

★ BeamLLM: Vision-Empowered mmWave Beam Prediction with Large Language Models

Can Zheng, Jiguang He, Guofa Cai, Zitong Yu, Chung G. Kang

In this paper, we propose BeamLLM, a vision-aided millimeter-wave (mmWave) beam prediction framework leveraging large language models (LLMs) to address the challenges of high training overhead and latency in mmWave communication systems. By combining computer vision (CV) with LLMs' cross-modal reasoning capabilities, the framework extracts user equipment (UE) positional features from RGB images and aligns visual-temporal features with LLMs' semantic space through reprogramming techniques. Evaluated on a realistic vehicle-to-infrastructure (V2I) scenario, the proposed method achieves 61.01% top-1 accuracy and 97.39% top-3 accuracy in standard prediction tasks, significantly outperforming traditional deep learning models. In few-shot prediction scenarios, the performance degradation is limited to 12.56% (top-1) and 5.55% (top-3) from time sample 1 to 10, demonstrating superior prediction capability.

comment: 6 pages, 7 figures, conference

★ Langevin Monte-Carlo Provably Learns Depth Two Neural Nets at Any Size and Data

Dibyakanti Kumar, Samyak Jha, Anirbit Mukherjee

In this work, we will establish that the Langevin Monte-Carlo algorithm can learn depth-2 neural nets of any size and for any data and we give non-asymptotic convergence rates for it. We achieve this via showing that under Total Variation distance and q-Renyi divergence, the iterates of Langevin Monte Carlo converge to the Gibbs distribution of Frobenius norm regularized losses for any of these nets, when using smooth activations and in both classification and regression settings. Most critically, the amount of regularization needed for our results is independent of the size of the net. The key observation of ours is that two layer neural loss functions can always be regularized by a constant amount such that they satisfy the Villani conditions, and thus their Gibbs measures satisfy a Poincare inequality.

★ Improving Medical Waste Classification with Hybrid Capsule Networks

Bennet van den Broek, Javad Pourmostafa Roshan Sharami

The improper disposal and mismanagement of medical waste pose severe environmental and public health risks, contributing to greenhouse gas emissions and the spread of infectious diseases. Efficient and accurate medical waste classification is crucial for mitigating these risks. We explore the integration of capsule networks with a pretrained DenseNet model to improve medical waste classification. To the best of our knowledge, capsule networks have not yet been applied to this task, making this study the first to assess their effectiveness. A diverse dataset of medical waste images collected from multiple public sources, is used to evaluate three model configurations: (1) a pretrained DenseNet model as a baseline, (2) a pretrained DenseNet with frozen layers combined with a capsule network, and (3) a pretrained DenseNet with unfrozen layers combined with a capsule network. Experimental results demonstrate that incorporating capsule networks improves classification performance, with F1 scores increasing from 0.89 (baseline) to 0.92 (hybrid model with unfrozen layers). This highlights the potential of capsule networks to address the spatial limitations of traditional convolutional models and improve classification robustness. While the capsule-enhanced model demonstrated improved classification performance, direct comparisons with prior studies were challenging due to differences in dataset size and diversity. Previous studies relied on smaller, domain-specific datasets, which inherently yielded higher accuracy. In contrast, our study employs a significantly larger and more diverse dataset, leading to better generalization but introducing additional classification challenges. This highlights the trade-off between dataset complexity and model performance.

★ Towards Constraint-Based Adaptive Hypergraph Learning for Solving Vehicle Routing: An End-to-End Solution

Zhenwei Wang, Ruibin Bai, Tiehua Zhang

The application of learning based methods to vehicle routing problems has emerged as a pivotal area of research in combinatorial optimization. These problems are characterized by vast solution spaces and intricate constraints, making traditional approaches such as exact mathematical models or heuristic methods prone to high computational overhead or reliant on the design of complex heuristic operators to achieve optimal or near optimal solutions. Meanwhile, although some recent learning-based methods can produce good performance for VRP with straightforward constraint scenarios, they often fail to effectively handle hard constraints that are common in practice. This study introduces a novel end-to-end framework that combines constraint-oriented hypergraphs with reinforcement learning to address vehicle routing problems. A central innovation of this work is the development of a constraint-oriented dynamic hyperedge reconstruction strategy within an encoder, which significantly enhances hypergraph representation learning. Additionally, the decoder leverages a double-pointer attention mechanism to iteratively generate solutions. The proposed model is trained by incorporating asynchronous parameter updates informed by hypergraph constraints and optimizing a dual loss function comprising constraint loss and policy gradient loss. The experiment results on benchmark datasets demonstrate that the proposed approach not only eliminates the need for sophisticated heuristic operators but also achieves substantial improvements in solution quality.

★ dFLMoE: Decentralized Federated Learning via Mixture of Experts for Medical Data Analysis

Luyuan Xie, Tianyu Luan, Wenyuan Cai, Guochen Yan, Zhaoyu Chen, Nan Xi, Yuejian Fang, Qingni Shen, Zhonghai Wu, Junsong Yuan

Federated learning has wide applications in the medical field. It enables knowledge sharing among different healthcare institutes while protecting patients' privacy. However, existing federated learning systems are typically centralized, requiring clients to upload client-specific knowledge to a central server for aggregation. This centralized approach would integrate the knowledge from each client into a centralized server, and the knowledge would be already undermined during the centralized integration before it reaches back to each client. Besides, the centralized approach also creates a dependency on the central server, which may affect training stability if the server malfunctions or connections are unstable. To address these issues, we propose a decentralized federated learning framework named dFLMoE. In our framework, clients directly exchange lightweight head models with each other. After exchanging, each client treats both local and received head models as individual experts, and utilizes a client-specific Mixture of Experts (MoE) approach to make collective decisions. This design not only reduces the knowledge damage with client-specific aggregations but also removes the dependency on the central server to enhance the robustness of the framework. We validate our framework on multiple medical tasks, demonstrating that our method evidently outperforms state-of-the-art approaches under both model homogeneity and heterogeneity settings.

★ Understanding the Logical Capabilities of Large Language Models via Out-of-Context Representation Learning

Jonathan Shaki, Emanuele La Malfa, Michael Wooldridge, Sarit Kraus

We study the capabilities of Large Language Models (LLM) on binary relations, a ubiquitous concept in math employed in most reasoning, math and logic benchmarks. This work focuses on equality, inequality, and inclusion, along with the properties they satisfy, such as ir/reflexivity, a/symmetry, transitivity, and logical complexity (e.g., number of reasoning ``hops''). We propose an alternative to in-context learning that trains only the representations of newly introduced tokens, namely out-of-context representation learning. This method mitigates linguistic biases already present in a model and, differently from in-context learning, does not rely on external information or illustrations. We argue out-of-context representation learning as a better alternative to in-context learning and fine-tuning to evaluate the capabilities of LLMs on logic tasks that are the building blocks of more complex reasoning benchmarks.

★ Architecture-Aware Minimization (A$^2$M): How to Find Flat Minima in Neural Architecture Search

Matteo Gambella, Fabrizio Pittorino, Manuel Roveri

Neural Architecture Search (NAS) has become an essential tool for designing effective and efficient neural networks. In this paper, we investigate the geometric properties of neural architecture spaces commonly used in differentiable NAS methods, specifically NAS-Bench-201 and DARTS. By defining flatness metrics such as neighborhoods and loss barriers along paths in architecture space, we reveal locality and flatness characteristics analogous to the well-known properties of neural network loss landscapes in weight space. In particular, we find that highly accurate architectures cluster together in flat regions, while suboptimal architectures remain isolated, unveiling the detailed geometrical structure of the architecture search landscape. Building on these insights, we propose Architecture-Aware Minimization (A$^2$M), a novel analytically derived algorithmic framework that explicitly biases, for the first time, the gradient of differentiable NAS methods towards flat minima in architecture space. A$^2$M consistently improves generalization over state-of-the-art DARTS-based algorithms on benchmark datasets including CIFAR-10, CIFAR-100, and ImageNet16-120, across both NAS-Bench-201 and DARTS search spaces. Notably, A$^2$M is able to increase the test accuracy, on average across different differentiable NAS methods, by +3.60\% on CIFAR-10, +4.60\% on CIFAR-100, and +3.64\% on ImageNet16-120, demonstrating its superior effectiveness in practice. A$^2$M can be easily integrated into existing differentiable NAS frameworks, offering a versatile tool for future research and applications in automated machine learning. We open-source our code at https://github.com/AI-Tech-Research-Lab/AsquaredM.

comment: 22 pages, 11 figures, 3 tables

★ Multi-objective Good Arm Identification with Bandit Feedback

Xuanke Jiang, Kohei Hatano, Eiji Takimoto

We consider a good arm identification problem in a stochastic bandit setting with multi-objectives, where each arm $i\in[K]$ is associated with $M$ distributions $\mathcal{D}_i^{(1)}, \ldots, \mathcal{D}_i^{(M)}$. For each round $t$, the player/algorithm pulls one arm $i_t$ and receives a vector feedback, where each component $m$ is sampled according to $\mathcal{D}_i^{(m)}$. The target is twofold, one is finding one arm whose means are larger than the predefined thresholds $\xi_1,\ldots,\xi_M$ with a confidence bound $\delta$ and an accuracy rate $\epsilon$ with a bounded sample complexity, the other is output $\bot$ to indicate no such arm exists. We propose an algorithm with a sample complexity bound. When $M=1$ and $\epsilon = 0$, our bound is the same as the one given in the previous work when and novel bounds for $M > 1$. The proposed algorithm attains better numerical performance than other baselines in the experiments on synthetic and real datasets.

★ Subgroup Performance Analysis in Hidden Stratifications

Alceu Bissoto, Trung-Dung Hoang, Tim Flühmann, Susu Sun, Christian F. Baumgartner, Lisa M. Koch

Machine learning (ML) models may suffer from significant performance disparities between patient groups. Identifying such disparities by monitoring performance at a granular level is crucial for safely deploying ML to each patient. Traditional subgroup analysis based on metadata can expose performance disparities only if the available metadata (e.g., patient sex) sufficiently reflects the main reasons for performance variability, which is not common. Subgroup discovery techniques that identify cohesive subgroups based on learned feature representations appear as a potential solution: They could expose hidden stratifications and provide more granular subgroup performance reports. However, subgroup discovery is challenging to evaluate even as a standalone task, as ground truth stratification labels do not exist in real data. Subgroup discovery has thus neither been applied nor evaluated for the application of subgroup performance monitoring. Here, we apply subgroup discovery for performance monitoring in chest x-ray and skin lesion classification. We propose novel evaluation strategies and show that a simplified subgroup discovery method without access to classification labels or metadata can expose larger performance disparities than traditional metadata-based subgroup analysis. We provide the first compelling evidence that subgroup discovery can serve as an important tool for comprehensive performance validation and monitoring of trustworthy AI in medicine.

comment: Under review

★ Probabilistic Forecasting via Autoregressive Flow Matching

Ahmed El-Gazzar, Marcel van Gerven

In this work, we propose FlowTime, a generative model for probabilistic forecasting of multivariate timeseries data. Given historical measurements and optional future covariates, we formulate forecasting as sampling from a learned conditional distribution over future trajectories. Specifically, we decompose the joint distribution of future observations into a sequence of conditional densities, each modeled via a shared flow that transforms a simple base distribution into the next observation distribution, conditioned on observed covariates. To achieve this, we leverage the flow matching (FM) framework, enabling scalable and simulation-free learning of these transformations. By combining this factorization with the FM objective, FlowTime retains the benefits of autoregressive models -- including strong extrapolation performance, compact model size, and well-calibrated uncertainty estimates -- while also capturing complex multi-modal conditional distributions, as seen in modern transport-based generative models. We demonstrate the effectiveness of FlowTime on multiple dynamical systems and real-world forecasting tasks.

★ A Multimodal Fusion Model Leveraging MLP Mixer and Handcrafted Features-based Deep Learning Networks for Facial Palsy Detection PAKDD 2025

Heng Yim Nicole Oo, Min Hun Lee, Jeong Hoon Lim

Algorithmic detection of facial palsy offers the potential to improve current practices, which usually involve labor-intensive and subjective assessments by clinicians. In this paper, we present a multimodal fusion-based deep learning model that utilizes an MLP mixer-based model to process unstructured data (i.e. RGB images or images with facial line segments) and a feed-forward neural network to process structured data (i.e. facial landmark coordinates, features of facial expressions, or handcrafted features) for detecting facial palsy. We then contribute to a study to analyze the effect of different data modalities and the benefits of a multimodal fusion-based approach using videos of 20 facial palsy patients and 20 healthy subjects. Our multimodal fusion model achieved 96.00 F1, which is significantly higher than the feed-forward neural network trained on handcrafted features alone (82.80 F1) and an MLP mixer-based model trained on raw RGB images (89.00 F1).

comment: PAKDD 2025. arXiv admin note: text overlap with arXiv:2405.16496

★ LUMOS: Language-Conditioned Imitation Learning with World Models ICRA

Iman Nematollahi, Branton DeMoss, Akshay L Chandra, Nick Hawes, Wolfram Burgard, Ingmar Posner

We introduce LUMOS, a language-conditioned multi-task imitation learning framework for robotics. LUMOS learns skills by practicing them over many long-horizon rollouts in the latent space of a learned world model and transfers these skills zero-shot to a real robot. By learning on-policy in the latent space of the learned world model, our algorithm mitigates policy-induced distribution shift which most offline imitation learning methods suffer from. LUMOS learns from unstructured play data with fewer than 1% hindsight language annotations but is steerable with language commands at test time. We achieve this coherent long-horizon performance by combining latent planning with both image- and language-based hindsight goal relabeling during training, and by optimizing an intrinsic reward defined in the latent space of the world model over multiple time steps, effectively reducing covariate shift. In experiments on the difficult long-horizon CALVIN benchmark, LUMOS outperforms prior learning-based methods with comparable approaches on chained multi-task evaluations. To the best of our knowledge, we are the first to learn a language-conditioned continuous visuomotor control for a real-world robot within an offline world model. Videos, dataset and code are available at http://lumos.cs.uni-freiburg.de.

comment: Accepted at the 2025 IEEE International Conference on Robotics and Automation (ICRA)

★ BioSerenity-E1: a self-supervised EEG model for medical applications

Ruggero G. Bettinardi, Mohamed Rahmouni, Ulysse Gimenez

Electroencephalography (EEG) serves as an essential diagnostic tool in neurology; however, its accurate manual interpretation is a time-intensive process that demands highly specialized expertise, which remains relatively scarce and not consistently accessible. To address these limitations, the implementation of automated pre-screening and analysis systems for EEG data holds considerable promise. Advances in self-supervised learning made it possible to pre-train complex deep learning architectures on large volumes of unlabeled EEG data to learn generalizable representations, that can later be used to enhance performance on multiple tasks while needing less downstream data. In the present paper, we introduce BioSerenity-E1, the first of a family of self-supervised foundation models for clinical EEG applications that combines spectral tokenization with masked prediction to achieve state-of-the-art performance across relevant diagnostic tasks. The two-phase self-supervised pretraining framework initially acquires compressed EEG representations via a transformer-based VQ-VAE architecture designed to reconstruct log-multitaper spectral projections, then implements extensive (70% block) masked token prediction to force the model to learn complex spatiotemporal dependencies in EEG signals. BioSerenity-E1 achieves strong performance across three clinical tasks, either in line or above state-of-the-art methods: seizure detection (AUROC = 0.926, Sensitivity = 0.909), normal/abnormal classification (AUPRC = 0.970 on proprietary data; 0.910 on TUH-Abnormal), and multiclass pathology differentiation on unbalanced data (Weighted F1 = 0.730). The utility of BioSerenity-E1 is further confirmed in low-data regimes scenarios, showing clear improvements in AUPRC (from +2% to 17%) when trained on less than 10% of the available data.

★ ConceptGuard: Continual Personalized Text-to-Image Generation with Forgetting and Confusion Mitigation CVPR 2025

Zirun Guo, Tao Jin

Diffusion customization methods have achieved impressive results with only a minimal number of user-provided images. However, existing approaches customize concepts collectively, whereas real-world applications often require sequential concept integration. This sequential nature can lead to catastrophic forgetting, where previously learned concepts are lost. In this paper, we investigate concept forgetting and concept confusion in the continual customization. To tackle these challenges, we present ConceptGuard, a comprehensive approach that combines shift embedding, concept-binding prompts and memory preservation regularization, supplemented by a priority queue which can adaptively update the importance and occurrence order of different concepts. These strategies can dynamically update, unbind and learn the relationship of the previous concepts, thus alleviating concept forgetting and confusion. Through comprehensive experiments, we show that our approach outperforms all the baseline methods consistently and significantly in both quantitative and qualitative analyses.

comment: Accepted at CVPR 2025

★ Safe exploration in reproducing kernel Hilbert spaces AISTATS 2025

Abdullah Tokmak, Kiran G. Krishnan, Thomas B. Schön, Dominik Baumann

Popular safe Bayesian optimization (BO) algorithms learn control policies for safety-critical systems in unknown environments. However, most algorithms make a smoothness assumption, which is encoded by a known bounded norm in a reproducing kernel Hilbert space (RKHS). The RKHS is a potentially infinite-dimensional space, and it remains unclear how to reliably obtain the RKHS norm of an unknown function. In this work, we propose a safe BO algorithm capable of estimating the RKHS norm from data. We provide statistical guarantees on the RKHS norm estimation, integrate the estimated RKHS norm into existing confidence intervals and show that we retain theoretical guarantees, and prove safety of the resulting safe BO algorithm. We apply our algorithm to safely optimize reinforcement learning policies on physics simulators and on a real inverted pendulum, demonstrating improved performance, safety, and scalability compared to the state-of-the-art.

comment: Accepted to AISTATS 2025

★ Mirror Online Conformal Prediction with Intermittent Feedback

Bowen Wang, Matteo Zecchin, Osvaldo Simeone

Online conformal prediction enables the runtime calibration of a pre-trained artificial intelligence model using feedback on its performance. Calibration is achieved through set predictions that are updated via online rules so as to ensure long-term coverage guarantees. While recent research has demonstrated the benefits of incorporating prior knowledge into the calibration process, this has come at the cost of replacing coverage guarantees with less tangible regret guarantees based on the quantile loss. This work introduces intermittent mirror online conformal prediction (IM-OCP), a novel runtime calibration framework that integrates prior knowledge, while maintaining long-term coverage and achieving sub-linear regret. IM-OCP features closed-form updates with minimal memory complexity, and is designed to operate under potentially intermittent feedback.

★ Characterizing Nonlinear Dynamics via Smooth Prototype Equivalences

Roy Friedman, Noa Moriel, Matthew Ricci, Guy Pelc, Yair Weiss, Mor Nitzan

Characterizing dynamical systems given limited measurements is a common challenge throughout the physical and biological sciences. However, this task is challenging, especially due to transient variability in systems with equivalent long-term dynamics. We address this by introducing smooth prototype equivalences (SPE), a framework that fits a diffeomorphism using normalizing flows to distinct prototypes - simplified dynamical systems that define equivalence classes of behavior. SPE enables classification by comparing the deformation loss of the observed sparse, high-dimensional measurements to the prototype dynamics. Furthermore, our approach enables estimation of the invariant sets of the observed dynamics through the learned mapping from prototype space to data space. Our method outperforms existing techniques in the classification of oscillatory systems and can efficiently identify invariant structures like limit cycles and fixed points in an equation-free manner, even when only a small, noisy subset of the phase space is observed. Finally, we show how our method can be used for the detection of biological processes like the cell cycle trajectory from high-dimensional single-cell gene expression data.

comment: 9 pages, 6 figures

★ Generative Binary Memory: Pseudo-Replay Class-Incremental Learning on Binarized Embeddings

Yanis Basso-Bert, Anca Molnos, Romain Lemaire, William Guicquero, Antoine Dupret

In dynamic environments where new concepts continuously emerge, Deep Neural Networks (DNNs) must adapt by learning new classes while retaining previously acquired ones. This challenge is addressed by Class-Incremental Learning (CIL). This paper introduces Generative Binary Memory (GBM), a novel CIL pseudo-replay approach which generates synthetic binary pseudo-exemplars. Relying on Bernoulli Mixture Models (BMMs), GBM effectively models the multi-modal characteristics of class distributions, in a latent, binary space. With a specifically-designed feature binarizer, our approach applies to any conventional DNN. GBM also natively supports Binary Neural Networks (BNNs) for highly-constrained model sizes in embedded systems. The experimental results demonstrate that GBM achieves higher than state-of-the-art average accuracy on CIFAR100 (+2.9%) and TinyImageNet (+1.5%) for a ResNet-18 equipped with our binarizer. GBM also outperforms emerging CIL methods for BNNs, with +3.1% in final accuracy and x4.7 memory reduction, on CORE50.

★ Collaborative Speculative Inference for Efficient LLM Inference Serving

Luyao Gao, Jianchun Liu, Hongli Xu, Liusheng Huang

Speculative inference is a promising paradigm employing small speculative models (SSMs) as drafters to generate draft tokens, which are subsequently verified in parallel by the target large language model (LLM). This approach enhances the efficiency of inference serving by reducing LLM inference latency and costs while preserving generation quality. However, existing speculative methods face critical challenges, including inefficient resource utilization and limited draft acceptance, which constrain their scalability and overall effectiveness. To overcome these obstacles, we present CoSine, a novel speculative inference system that decouples sequential speculative decoding from parallel verification, enabling efficient collaboration among multiple nodes. Specifically, CoSine routes inference requests to specialized drafters based on their expertise and incorporates a confidence-based token fusion mechanism to synthesize outputs from cooperating drafters, ensuring high-quality draft generation. Additionally, CoSine dynamically orchestrates the execution of speculative decoding and verification in a pipelined manner, employing batch scheduling to selectively group requests and adaptive speculation control to minimize idle periods. By optimizing parallel workflows through heterogeneous node collaboration, CoSine balances draft generation and verification throughput in real-time, thereby maximizing resource utilization. Experimental results demonstrate that CoSine achieves superior performance compared to state-of-the-art speculative approaches. Notably, with equivalent resource costs, CoSine achieves up to a 23.2% decrease in latency and a 32.5% increase in throughput compared to baseline methods.

★ Enhance Exploration in Safe Reinforcement Learning with Contrastive Representation Learning

Duc Kien Doan, Bang Giang Le, Viet Cuong Ta

In safe reinforcement learning, agent needs to balance between exploration actions and safety constraints. Following this paradigm, domain transfer approaches learn a prior Q-function from the related environments to prevent unsafe actions. However, because of the large number of false positives, some safe actions are never executed, leading to inadequate exploration in sparse-reward environments. In this work, we aim to learn an efficient state representation to balance the exploration and safety-prefer action in a sparse-reward environment. Firstly, the image input is mapped to latent representation by an auto-encoder. A further contrastive learning objective is employed to distinguish safe and unsafe states. In the learning phase, the latent distance is used to construct an additional safety check, which allows the agent to bias the exploration if it visits an unsafe state. To verify the effectiveness of our method, the experiment is carried out in three navigation-based MiniGrid environments. The result highlights that our method can explore the environment better while maintaining a good balance between safety and efficiency.

comment: Accepted at ACIIDS 2025

★ Capturing Semantic Flow of ML-based Systems

Shin Yoo, Robert Feldt, Somin Kim, Naryeong Kim

ML-based systems are software systems that incorporates machine learning components such as Deep Neural Networks (DNNs) or Large Language Models (LLMs). While such systems enable advanced features such as high performance computer vision, natural language processing, and code generation, their internal behaviour remain largely opaque to traditional dynamic analysis such as testing: existing analysis typically concern only what is observable from the outside, such as input similarity or class label changes. We propose semantic flow, a concept designed to capture the internal behaviour of ML-based system and to provide a platform for traditional dynamic analysis techniques to be adapted to. Semantic flow combines the idea of control flow with internal states taken from executions of ML-based systems, such as activation values of a specific layer in a DNN, or embeddings of LLM responses at a specific inference step of LLM agents. The resulting representation, summarised as semantic flow graphs, can capture internal decisions that are not explicitly represented in the traditional control flow of ML-based systems. We propose the idea of semantic flow, introduce two examples using a DNN and an LLM agent, and finally sketch its properties and how it can be used to adapt existing dynamic analysis techniques for use in ML-based software systems.

★ Nash Equilibrium Constrained Auto-bidding With Bi-level Reinforcement Learning

Zhiyu Mou, Miao Xu, Rongquan Bai, Zhuoran Yang, Chuan Yu, Jian Xu, Bo Zheng

Many online advertising platforms provide advertisers with auto-bidding services to enhance their advertising performance. However, most existing auto-bidding algorithms fail to accurately capture the auto-bidding problem formulation that the platform truly faces, let alone solve it. Actually, we argue that the platform should try to help optimize each advertiser's performance to the greatest extent -- which makes $\epsilon$-Nash Equilibrium ($\epsilon$-NE) a necessary solution concept -- while maximizing the social welfare of all the advertisers for the platform's long-term value. Based on this, we introduce the \emph{Nash-Equilibrium Constrained Bidding} (NCB), a new formulation of the auto-bidding problem from the platform's perspective. Specifically, it aims to maximize the social welfare of all advertisers under the $\epsilon$-NE constraint. However, the NCB problem presents significant challenges due to its constrained bi-level structure and the typically large number of advertisers involved. To address these challenges, we propose a \emph{Bi-level Policy Gradient} (BPG) framework with theoretical guarantees. Notably, its computational complexity is independent of the number of advertisers, and the associated gradients are straightforward to compute. Extensive simulated and real-world experiments validate the effectiveness of the BPG framework.

★ Wikipedia is Not a Dictionary, Delete! Text Classification as a Proxy for Analysing Wiki Deletion Discussions

Hsuvas Borkakoty, Luis Espinosa-Anke

Automated content moderation for collaborative knowledge hubs like Wikipedia or Wikidata is an important yet challenging task due to multiple factors. In this paper, we construct a database of discussions happening around articles marked for deletion in several Wikis and in three languages, which we then use to evaluate a range of LMs on different tasks (from predicting the outcome of the discussion to identifying the implicit policy an individual comment might be pointing to). Our results reveal, among others, that discussions leading to deletion are easier to predict, and that, surprisingly, self-produced tags (keep, delete or redirect) don't always help guiding the classifiers, presumably because of users' hesitation or deliberation within comments.

comment: Accepted to WNUT-2025

★ PyGDA: A Python Library for Graph Domain Adaptation

Zhen Zhang, Meihan Liu, Bingsheng He

Graph domain adaptation has emerged as a promising approach to facilitate knowledge transfer across different domains. Recently, numerous models have been proposed to enhance their generalization capabilities in this field. However, there is still no unified library that brings together existing techniques and simplifies their implementation. To fill this gap, we introduce PyGDA, an open-source Python library tailored for graph domain adaptation. As the first comprehensive library in this area, PyGDA covers more than 20 widely used graph domain adaptation methods together with different types of graph datasets. Specifically, PyGDA offers modular components, enabling users to seamlessly build custom models with a variety of commonly used utility functions. To handle large-scale graphs, PyGDA includes support for features such as sampling and mini-batch processing, ensuring efficient computation. In addition, PyGDA also includes comprehensive performance benchmarks and well-documented user-friendly API for both researchers and practitioners. To foster convenient accessibility, PyGDA is released under the MIT license at https://github.com/pygda-team/pygda, and the API documentation is https://pygda.readthedocs.io/en/stable/.

comment: Under Review

★ HyperArm Bandit Optimization: A Novel approach to Hyperparameter Optimization and an Analysis of Bandit Algorithms in Stochastic and Adversarial Settings

Samih Karroum, Saad Mazhar

This paper explores the application of bandit algorithms in both stochastic and adversarial settings, with a focus on theoretical analysis and practical applications. The study begins by introducing bandit problems, distinguishing between stochastic and adversarial variants, and examining key algorithms such as Explore-Then-Commit (ETC), Upper Confidence Bound (UCB), and Exponential-Weight Algorithm for Exploration and Exploitation (EXP3). Theoretical regret bounds are analyzed to compare the performance of these algorithms. The paper then introduces a novel framework, HyperArm Bandit Optimization (HABO), which applies EXP3 to hyperparameter tuning in machine learning models. Unlike traditional methods that treat entire configurations as arms, HABO treats individual hyperparameters as super-arms, and its potential configurations as sub-arms, enabling dynamic resource allocation and efficient exploration. Experimental results demonstrate HABO's effectiveness in classification and regression tasks, outperforming Bayesian Optimization in terms of computational efficiency and accuracy. The paper concludes with insights into the convergence guarantees of HABO and its potential for scalable and robust hyperparameter optimization.

comment: 41 pages, 9 figures

★ Robust Learning-Based Sparse Recovery for Device Activity Detection in Grant-Free Random Access Cell-Free Massive MIMO: Enhancing Resilience to Impairments

Ali Elkeshawy, Haifa Fares, Amor Nafkha

Massive MIMO is considered a key enabler to support massive machine-type communication (mMTC). While massive access schemes have been extensively analyzed for co-located massive MIMO arrays, this paper explores activity detection in grant-free random access for mMTC within the context of cell-free massive MIMO systems, employing distributed antenna arrays. This sparse support recovery of device activity status is performed by a finite cluster of access points (APs) from a large number of geographically distributed APs collaborating to serve a larger number of devices. Active devices transmit non-orthogonal pilot sequences to APs, which forward the received signals to a central processing unit (CPU) for collaborative activity detection. This paper proposes a simple and efficient data-driven algorithm tailored for device activity detection, implemented centrally at the CPU. Furthermore, the study assesses the algorithm's robustness to input perturbations and examines the effects of adopting fixed-point representation on its performance.

★ Numerically robust Gaussian state estimation with singular observation noise

Nicholas Krämer, Filip Tronarp

This article proposes numerically robust algorithms for Gaussian state estimation with singular observation noise. Our approach combines a series of basis changes with Bayes' rule, transforming the singular estimation problem into a nonsingular one with reduced state dimension. In addition to ensuring low runtime and numerical stability, our proposal facilitates marginal-likelihood computations and Gauss-Markov representations of the posterior process. We analyse the proposed method's computational savings and numerical robustness and validate our findings in a series of simulations.

★ Climate land use and other drivers impacts on island ecosystem services: a global review

Aristides Moustakas, Shiri Zemah-Shamir, Mirela Tase, Savvas Zotos, Nazli Demirel, Christos Zoumides, Irene Christoforidi, Turgay Dindaroglu, Tamer Albayrak, Cigdem Kaptan Ayhan, Mauro Fois, Paraskevi Manolaki, Attila D. Sandor, Ina Sieber, Valentini Stamatiadou, Elli Tzirkalli, Ioannis N. Vogiatzakis, Ziv Zemah-Shamir, George Zittis

Islands are diversity hotspots and vulnerable to environmental degradation, climate variations, land use changes and societal crises. These factors can exhibit interactive impacts on ecosystem services. The study reviewed a large number of papers on the climate change-islands-ecosystem services topic worldwide. Potential inclusion of land use changes and other drivers of impacts on ecosystem services were sequentially also recorded. The study sought to investigate the impacts of climate change, land use change, and other non-climatic driver changes on island ecosystem services. Explanatory variables examined were divided into two categories: environmental variables and methodological ones. Environmental variables include sea zone geographic location, ecosystem, ecosystem services, climate, land use, other driver variables, Methodological variables include consideration of policy interventions, uncertainty assessment, cumulative effects of climate change, synergistic effects of climate change with land use change and other anthropogenic and environmental drivers, and the diversity of variables used in the analysis. Machine learning and statistical methods were used to analyze their effects on island ecosystem services. Negative climate change impacts on ecosystem services are better quantified by land use change or other non-climatic driver variables than by climate variables. The synergy of land use together with climate changes is modulating the impact outcome and critical for a better impact assessment. Analyzed together, there is little evidence of more pronounced for a specific sea zone, ecosystem, or ecosystem service. Climate change impacts may be underestimated due to the use of a single climate variable deployed in most studies. Policy interventions exhibit low classification accuracy in quantifying impacts indicating insufficient efficacy or integration in the studies.

comment: Article published in the journal: Science of the Total Environment. Free author's version

★ Resource efficient data transmission on animals based on machine learning

Wilhelm Kerle-Malcharek, Karsten Klein, Martin Wikelski, Falk Schreiber, Timm A. Wild

Bio-loggers, electronic devices used to track animal behaviour through various sensors, have become essential in wildlife research. Despite continuous improvements in their capabilities, bio-loggers still face significant limitations in storage, processing, and data transmission due to the constraints of size and weight, which are necessary to avoid disturbing the animals. This study aims to explore how selective data transmission, guided by machine learning, can reduce the energy consumption of bio-loggers, thereby extending their operational lifespan without requiring hardware modifications.

comment: Submitted to Scientific Reports but not published, 23 pages, 5 figures, 3 tables

★ Targeted Data Poisoning for Black-Box Audio Datasets Ownership Verification ICASSP 2025

Wassim Bouaziz, El-Mahdi El-Mhamdi, Nicolas Usunier

Protecting the use of audio datasets is a major concern for data owners, particularly with the recent rise of audio deep learning models. While watermarks can be used to protect the data itself, they do not allow to identify a deep learning model trained on a protected dataset. In this paper, we adapt to audio data the recently introduced data taggants approach. Data taggants is a method to verify if a neural network was trained on a protected image dataset with top-$k$ predictions access to the model only. This method relies on a targeted data poisoning scheme by discreetly altering a small fraction (1%) of the dataset as to induce a harmless behavior on out-of-distribution data called keys. We evaluate our method on the Speechcommands and the ESC50 datasets and state of the art transformer models, and show that we can detect the use of the dataset with high confidence without loss of performance. We also show the robustness of our method against common data augmentation techniques, making it a practical method to protect audio datasets.

comment: Published at ICASSP 2025, 5 pages, 7 figures

★ AMR-Transformer: Enabling Efficient Long-range Interaction for Complex Neural Fluid Simulation

Zeyi Xu, Jinfan Liu, Kuangxu Chen, Ye Chen, Zhangli Hu, Bingbing Ni

Accurately and efficiently simulating complex fluid dynamics is a challenging task that has traditionally relied on computationally intensive methods. Neural network-based approaches, such as convolutional and graph neural networks, have partially alleviated this burden by enabling efficient local feature extraction. However, they struggle to capture long-range dependencies due to limited receptive fields, and Transformer-based models, while providing global context, incur prohibitive computational costs. To tackle these challenges, we propose AMR-Transformer, an efficient and accurate neural CFD-solving pipeline that integrates a novel adaptive mesh refinement scheme with a Navier-Stokes constraint-aware fast pruning module. This design encourages long-range interactions between simulation cells and facilitates the modeling of global fluid wave patterns, such as turbulence and shockwaves. Experiments show that our approach achieves significant gains in efficiency while preserving critical details, making it suitable for high-resolution physical simulations with long-range dependencies. On CFDBench, PDEBench and a new shockwave dataset, our pipeline demonstrates up to an order-of-magnitude improvement in accuracy over baseline models. Additionally, compared to ViT, our approach achieves a reduction in FLOPs of up to 60 times.

★ PIMRL: Physics-Informed Multi-Scale Recurrent Learning for Spatiotemporal Prediction

Han Wan, Qi Wang, Hao Sun

Simulation of spatiotemporal systems governed by partial differential equations is widely applied in fields such as biology, chemistry, aerospace dynamics, and meteorology. Traditional numerical methods incur high computational costs due to the requirement of small time steps for accurate predictions. While machine learning has reduced these costs, long-term predictions remain challenged by error accumulation, particularly in scenarios with insufficient data or varying time scales, where stability and accuracy are compromised. Existing methods often neglect the effective utilization of multi-scale data, leading to suboptimal robustness in predictions. To address these issues, we propose a novel multi-scale learning framework, namely, the Physics-Informed Multi-Scale Recurrent Learning (PIMRL), to effectively leverage multi-scale data for spatiotemporal dynamics prediction. The PIMRL framework comprises two modules: the micro-scale module embeds physical knowledge into neural networks via pretraining, and the macro-scale module adopts a data-driven approach to learn the temporal evolution of physics in the latent space. Experimental results demonstrate that the PIMRL framework consistently achieves state-of-the-art performance across five benchmark datasets ranging from one to three dimensions, showing average improvements of over 9\% in both RMSE and MAE evaluation metrics, with maximum enhancements reaching up to 80%.

★ Numerical Error Analysis of Large Language Models

Stanislav Budzinskiy, Wenyi Fang, Longbin Zeng, Philipp Petersen

Large language models based on transformer architectures have become integral to state-of-the-art natural language processing applications. However, their training remains computationally expensive and exhibits instabilities, some of which are expected to be caused by finite-precision computations. We provide a theoretical analysis of the impact of round-off errors within the forward pass of a transformer architecture which yields fundamental bounds for these effects. In addition, we conduct a series of numerical experiments which demonstrate the practical relevance of our bounds. Our results yield concrete guidelines for choosing hyperparameters that mitigate round-off errors, leading to more robust and stable inference.

★ Spherical dimension

Bogdan Chornomaz, Shay Moran, Tom Waknine

We introduce and study the spherical dimension, a natural topological relaxation of the VC dimension that unifies several results in learning theory where topology plays a key role in the proofs. The spherical dimension is defined by extending the set of realizable datasets (used to define the VC dimension) to the continuous space of realizable distributions. In this space, a shattered set of size d (in the VC sense) is completed into a continuous object, specifically a d-dimensional sphere of realizable distributions. The spherical dimension is then defined as the dimension of the largest sphere in this space. Thus, the spherical dimension is at least the VC dimension. The spherical dimension serves as a common foundation for leveraging the Borsuk-Ulam theorem and related topological tools. We demonstrate the utility of the spherical dimension in diverse applications, including disambiguations of partial concept classes, reductions from classification to stochastic convex optimization, stability and replicability, and sample compression schemes. Perhaps surprisingly, we show that the open question posed by Alon, Hanneke, Holzman, and Moran (FOCS 2021) of whether there exist non-trivial disambiguations for halfspaces with margin is equivalent to the basic open question of whether the VC and spherical dimensions are finite together.

★ Flows on convex polytopes

Tomek Diederen, Nicola Zamboni

We present a framework for modeling complex, high-dimensional distributions on convex polytopes by leveraging recent advances in discrete and continuous normalizing flows on Riemannian manifolds. We show that any full-dimensional polytope is homeomorphic to a unit ball, and our approach harnesses flows defined on the ball, mapping them back to the original polytope. Furthermore, we introduce a strategy to construct flows when only the vertex representation of a polytope is available, employing maximum entropy barycentric coordinates and Aitchison geometry. Our experiments take inspiration from applications in metabolic flux analysis and demonstrate that our methods achieve competitive density estimation, sampling accuracy, as well as fast training and inference times.

★ Policy Teaching via Data Poisoning in Learning from Human Preferences AISTATS 2025

Andi Nika, Jonathan Nöther, Debmalya Mandal, Parameswaran Kamalaruban, Adish Singla, Goran Radanović

We study data poisoning attacks in learning from human preferences. More specifically, we consider the problem of teaching/enforcing a target policy $\pi^\dagger$ by synthesizing preference data. We seek to understand the susceptibility of different preference-based learning paradigms to poisoned preference data by analyzing the number of samples required by the attacker to enforce $\pi^\dagger$. We first propose a general data poisoning formulation in learning from human preferences and then study it for two popular paradigms, namely: (a) reinforcement learning from human feedback (RLHF) that operates by learning a reward model using preferences; (b) direct preference optimization (DPO) that directly optimizes policy using preferences. We conduct a theoretical analysis of the effectiveness of data poisoning in a setting where the attacker is allowed to augment a pre-existing dataset and also study its special case where the attacker can synthesize the entire preference dataset from scratch. As our main results, we provide lower/upper bounds on the number of samples required to enforce $\pi^\dagger$. Finally, we discuss the implications of our results in terms of the susceptibility of these learning paradigms under such data poisoning attacks.

comment: In AISTATS 2025

★ Assessing the validity of new paradigmatic complexity measures as criterial features for proficiency in L2 writings in English

Cyriel Mallart, Andrew Simpkin, Nicolas Ballier, Paula Lissón, Rémi Venant, Jen-Yu Li, Bernardo Stearns, Thomas Gaillat

This article addresses Second Language (L2) writing development through an investigation of new grammatical and structural complexity metrics. We explore the paradigmatic production in learner English by linking language functions to specific grammatical paradigms. Using the EFCAMDAT as a gold standard and a corpus of French learners as an external test set, we employ a supervised learning framework to operationalise and evaluate seven microsystems. We show that learner levels are associated with the seven microsystems (MS). Using ordinal regression modelling for evaluation, the results show that all MS are significant but yield a low impact if taken individually. However, their influence is shown to be impactful if taken as a group. These microsystems and their measurement method suggest that it is possible to use them as part of broader-purpose CALL systems focused on proficiency assessment.

★ Probability-Flow ODE in Infinite-Dimensional Function Spaces ICLR 2025

Kunwoo Na, Junghyun Lee, Se-Young Yun, Sungbin Lim

Recent advances in infinite-dimensional diffusion models have demonstrated their effectiveness and scalability in function generation tasks where the underlying structure is inherently infinite-dimensional. To accelerate inference in such models, we derive, for the first time, an analog of the probability-flow ODE (PF-ODE) in infinite-dimensional function spaces. Leveraging this newly formulated PF-ODE, we reduce the number of function evaluations while maintaining sample quality in function generation tasks, including applications to PDEs.

comment: 26 pages, 8 figures. Accepted to the ICLR 2025 DeLTa Workshop

★ Moss: Proxy Model-based Full-Weight Aggregation in Federated Learning with Heterogeneous Models

Yifeng Cai, Ziqi Zhang, Ding Li, Yao Guo, Xiangqun Chen

Modern Federated Learning (FL) has become increasingly essential for handling highly heterogeneous mobile devices. Current approaches adopt a partial model aggregation paradigm that leads to sub-optimal model accuracy and higher training overhead. In this paper, we challenge the prevailing notion of partial-model aggregation and propose a novel "full-weight aggregation" method named Moss, which aggregates all weights within heterogeneous models to preserve comprehensive knowledge. Evaluation across various applications demonstrates that Moss significantly accelerates training, reduces on-device training time and energy consumption, enhances accuracy, and minimizes network bandwidth utilization when compared to state-of-the-art baselines.

comment: Accepted by ACM IMWUT/Ubicomp 2025

★ Efficient Federated Fine-Tuning of Large Language Models with Layer Dropout

Shilong Wang, Jianchun Liu, Hongli Xu, Jiaming Yan, Xianjun Gao

Fine-tuning plays a crucial role in enabling pre-trained LLMs to evolve from general language comprehension to task-specific expertise. To preserve user data privacy, federated fine-tuning is often employed and has emerged as the de facto paradigm. However, federated fine-tuning is prohibitively inefficient due to the tension between LLM complexity and the resource constraint of end devices, incurring unaffordable fine-tuning overhead. Existing literature primarily utilizes parameter-efficient fine-tuning techniques to mitigate communication costs, yet computational and memory burdens continue to pose significant challenges for developers. This work proposes DropPEFT, an innovative federated PEFT framework that employs a novel stochastic transformer layer dropout method, enabling devices to deactivate a considerable fraction of LLMs layers during training, thereby eliminating the associated computational load and memory footprint. In DropPEFT, a key challenge is the proper configuration of dropout ratios for layers, as overhead and training performance are highly sensitive to this setting. To address this challenge, we adaptively assign optimal dropout-ratio configurations to devices through an exploration-exploitation strategy, achieving efficient and effective fine-tuning. Extensive experiments show that DropPEFT can achieve a 1.3-6.3\times speedup in model convergence and a 40%-67% reduction in memory footprint compared to state-of-the-art methods.

comment: 13 pages

★ Deep Learning for Time Series Forecasting: A Survey

Xiangjie Kong, Zhenghao Chen, Weiyao Liu, Kaili Ning, Lechao Zhang, Syauqie Muhammad Marier, Yichen Liu, Yuhao Chen, Feng Xia

Time series forecasting (TSF) has long been a crucial task in both industry and daily life. Most classical statistical models may have certain limitations when applied to practical scenarios in fields such as energy, healthcare, traffic, meteorology, and economics, especially when high accuracy is required. With the continuous development of deep learning, numerous new models have emerged in the field of time series forecasting in recent years. However, existing surveys have not provided a unified summary of the wide range of model architectures in this field, nor have they given detailed summaries of works in feature extraction and datasets. To address this gap, in this review, we comprehensively study the previous works and summarize the general paradigms of Deep Time Series Forecasting (DTSF) in terms of model architectures. Besides, we take an innovative approach by focusing on the composition of time series and systematically explain important feature extraction methods. Additionally, we provide an overall compilation of datasets from various domains in existing works. Finally, we systematically emphasize the significant challenges faced and future research directions in this field.

★ Robustness Tokens: Towards Adversarial Robustness of Transformers ECCV

Brian Pulfer, Yury Belousov, Slava Voloshynovskiy

Recently, large pre-trained foundation models have become widely adopted by machine learning practitioners for a multitude of tasks. Given that such models are publicly available, relying on their use as backbone models for downstream tasks might result in high vulnerability to adversarial attacks crafted with the same public model. In this work, we propose Robustness Tokens, a novel approach specific to the transformer architecture that fine-tunes a few additional private tokens with low computational requirements instead of tuning model parameters as done in traditional adversarial training. We show that Robustness Tokens make Vision Transformer models significantly more robust to white-box adversarial attacks while also retaining the original downstream performances.

comment: This paper has been accepted for publication at the European Conference on Computer Vision (ECCV), 2024

★ Data augmentation using diffusion models to enhance inverse Ising inference

Yechan Lim, Sangwon Lee, Junghyo Jo

Identifying model parameters from observed configurations poses a fundamental challenge in data science, especially with limited data. Recently, diffusion models have emerged as a novel paradigm in generative machine learning, capable of producing new samples that closely mimic observed data. These models learn the gradient of model probabilities, bypassing the need for cumbersome calculations of partition functions across all possible configurations. We explore whether diffusion models can enhance parameter inference by augmenting small datasets. Our findings demonstrate this potential through a synthetic task involving inverse Ising inference and a real-world application of reconstructing missing values in neural activity data. This study serves as a proof-of-concept for using diffusion models for data augmentation in physics-related problems, thereby opening new avenues in data science.

★ Multiplicative Learning

Han Kim, Hyungjoon Soh, Vipul Periwal, Junghyo Jo

Efficient training of artificial neural networks remains a key challenge in deep learning. Backpropagation (BP), the standard learning algorithm, relies on gradient descent and typically requires numerous iterations for convergence. In this study, we introduce Expectation Reflection (ER), a novel learning approach that updates weights multiplicatively based on the ratio of observed to predicted outputs. Unlike traditional methods, ER maintains consistency without requiring ad hoc loss functions or learning rate hyperparameters. We extend ER to multilayer networks and demonstrate its effectiveness in performing image classification tasks. Notably, ER achieves optimal weight updates in a single iteration. Additionally, we reinterpret ER as a modified form of gradient descent incorporating the inverse mapping of target propagation. These findings suggest that ER provides an efficient and scalable alternative for training neural networks.

★ Are Convex Optimization Curves Convex?

Guy Barzilai, Ohad Shamir

In this paper, we study when we might expect the optimization curve induced by gradient descent to be \emph{convex} -- precluding, for example, an initial plateau followed by a sharp decrease, making it difficult to decide when optimization should stop. Although such undesirable behavior can certainly occur when optimizing general functions, might it also occur in the benign and well-studied case of smooth convex functions? As far as we know, this question has not been tackled in previous work. We show, perhaps surprisingly, that the answer crucially depends on the choice of the step size. In particular, for the range of step sizes which are known to result in monotonic convergence to an optimal value, there is a regime where the optimization curve will be provably convex, and there is a regime where the curve can be non-convex. We also extend our results to gradient flow, and to the closely-related but different question of whether the gradient norm decreases monotonically.

comment: 13 pages

★ Gumiho: A Hybrid Architecture to Prioritize Early Tokens in Speculative Decoding

Jinze Li, Yixing Xu, Haiduo Huang, Xuanwu Yin, Dong Li, Edith C. H. Ngai, Emad Barsoum

Speculative decoding (SPD) aims to accelerate the auto-regressive token generation process of a target Large Language Model (LLM). Some approaches employ a draft model with multiple heads to predict a sequence of future tokens, where each head handles a token in the sequence. The target LLM verifies the predicted sequence and accepts aligned tokens, enabling efficient multi-token generation. However, existing methods assume that all tokens within a sequence are equally important, employing identical head structures and relying on a single-generation paradigm, either serial or parallel. To this end, we theoretically demonstrate that initial tokens in the draft sequence are more important than later ones. Building on this insight, we propose Gumiho, a hybrid model combining serial and parallel heads. Specifically, given the critical importance of early tokens, we employ a sophisticated Transformer architecture for the early draft heads in a serial configuration to improve accuracy. For later tokens, we utilize multiple lightweight MLP heads operating in parallel to enhance efficiency. By allocating more advanced model structures and longer running times to the early heads, Gumiho achieves improved overall performance. The experimental results demonstrate that our method outperforms existing approaches, fully validating its effectiveness.

comment: Paper under review

★ An Real-Sim-Real (RSR) Loop Framework for Generalizable Robotic Policy Transfer with Differentiable Simulation

Lu Shi, Yuxuan Xu, Shiyu Wang, Jinhao Huang, Wenhao Zhao, Yufei Jia, Zike Yan, Weibin Gu, Guyue Zhou

The sim-to-real gap remains a critical challenge in robotics, hindering the deployment of algorithms trained in simulation to real-world systems. This paper introduces a novel Real-Sim-Real (RSR) loop framework leveraging differentiable simulation to address this gap by iteratively refining simulation parameters, aligning them with real-world conditions, and enabling robust and efficient policy transfer. A key contribution of our work is the design of an informative cost function that encourages the collection of diverse and representative real-world data, minimizing bias and maximizing the utility of each data point for simulation refinement. This cost function integrates seamlessly into existing reinforcement learning algorithms (e.g., PPO, SAC) and ensures a balanced exploration of critical regions in the real domain. Furthermore, our approach is implemented on the versatile Mujoco MJX platform, and our framework is compatible with a wide range of robotic systems. Experimental results on several robotic manipulation tasks demonstrate that our method significantly reduces the sim-to-real gap, achieving high task performance and generalizability across diverse scenarios of both explicit and implicit environmental uncertainties.

★ Reconsidering Feature Structure Information and Latent Space Alignment in Partial Multi-label Feature Selection AAAI 25

Hanlin Pan, Kunpeng Liu, Wanfu Gao

The purpose of partial multi-label feature selection is to select the most representative feature subset, where the data comes from partial multi-label datasets that have label ambiguity issues. For label disambiguation, previous methods mainly focus on utilizing the information inside the labels and the relationship between the labels and features. However, the information existing in the feature space is rarely considered, especially in partial multi-label scenarios where the noises is considered to be concentrated in the label space while the feature information is correct. This paper proposes a method based on latent space alignment, which uses the information mined in feature space to disambiguate in latent space through the structural consistency between labels and features. In addition, previous methods overestimate the consistency of features and labels in the latent space after convergence. We comprehensively consider the similarity of latent space projections to feature space and label space, and propose new feature selection term. This method also significantly improves the positive label identification ability of the selected features. Comprehensive experiments demonstrate the superiority of the proposed method.

comment: 9pages,6 figures,accept at AAAI 25

★ IMPACT: Intelligent Motion Planning with Acceptable Contact Trajectories via Vision-Language Models

Yiyang Ling, Karan Owalekar, Oluwatobiloba Adesanya, Erdem Bıyık, Daniel Seita

Motion planning involves determining a sequence of robot configurations to reach a desired pose, subject to movement and safety constraints. Traditional motion planning finds collision-free paths, but this is overly restrictive in clutter, where it may not be possible for a robot to accomplish a task without contact. In addition, contacts range from relatively benign (e.g., brushing a soft pillow) to more dangerous (e.g., toppling a glass vase). Due to this diversity, it is difficult to characterize which contacts may be acceptable or unacceptable. In this paper, we propose IMPACT, a novel motion planning framework that uses Vision-Language Models (VLMs) to infer environment semantics, identifying which parts of the environment can best tolerate contact based on object properties and locations. Our approach uses the VLM's outputs to produce a dense 3D "cost map" that encodes contact tolerances and seamlessly integrates with standard motion planners. We perform experiments using 20 simulation and 10 real-world scenes and assess using task success rate, object displacements, and feedback from human evaluators. Our results over 3620 simulation and 200 real-world trials suggest that IMPACT enables efficient contact-rich motion planning in cluttered settings while outperforming alternative methods and ablations. Supplementary material is available at https://impact-planning.github.io/.

★ Improving Diffusion-based Inverse Algorithms under Few-Step Constraint via Learnable Linear Extrapolation

Jiawei Zhang, Ziyuan Liu, Leon Yan, Gen Li, Yuantao Gu

Diffusion models have demonstrated remarkable performance in modeling complex data priors, catalyzing their widespread adoption in solving various inverse problems. However, the inherently iterative nature of diffusion-based inverse algorithms often requires hundreds to thousands of steps, with performance degradation occurring under fewer steps which limits their practical applicability. While high-order diffusion ODE solvers have been extensively explored for efficient diffusion sampling without observations, their application to inverse problems remains underexplored due to the diverse forms of inverse algorithms and their need for repeated trajectory correction based on observations. To address this gap, we first introduce a canonical form that decomposes existing diffusion-based inverse algorithms into three modules to unify their analysis. Inspired by the linear subspace search strategy in the design of high-order diffusion ODE solvers, we propose the Learnable Linear Extrapolation (LLE) method, a lightweight approach that universally enhances the performance of any diffusion-based inverse algorithm that fits the proposed canonical form. Extensive experiments demonstrate consistent improvements of the proposed LLE method across multiple algorithms and tasks, indicating its potential for more efficient solutions and boosted performance of diffusion-based inverse algorithms with limited steps. Codes for reproducing our experiments are available at \href{https://github.com/weigerzan/LLE_inverse_problem}{https://github.com/weigerzan/LLE\_inverse\_problem}.

comment: preprint

♻ ★ Chain-of-Thought Reasoning In The Wild Is Not Always Faithful ICLR 25

Iván Arcuschin, Jett Janiak, Robert Krzyzanowski, Senthooran Rajamanoharan, Neel Nanda, Arthur Conmy

Chain-of-Thought (CoT) reasoning has significantly advanced state-of-the-art AI capabilities. However, recent studies have shown that CoT reasoning is not always faithful, i.e. CoT reasoning does not always reflect how models arrive at conclusions. So far, most of these studies have focused on unfaithfulness in unnatural contexts where an explicit bias has been introduced. In contrast, we show that unfaithful CoT can occur on realistic prompts with no artificial bias. Our results reveal non-negligible rates of several forms of unfaithful reasoning in frontier models: Sonnet 3.7 (16.3%), DeepSeek R1 (5.3%) and ChatGPT-4o (7.0%) all answer a notable proportion of question pairs unfaithfully. Specifically, we find that models rationalize their implicit biases in answers to binary questions ("implicit post-hoc rationalization"). For example, when separately presented with the questions "Is X bigger than Y?" and "Is Y bigger than X?", models sometimes produce superficially coherent arguments to justify answering Yes to both questions or No to both questions, despite such responses being logically contradictory. We also investigate restoration errors (Dziri et al., 2023), where models make and then silently correct errors in their reasoning, and unfaithful shortcuts, where models use clearly illogical reasoning to simplify solving problems in Putnam questions (a hard benchmark). Our findings raise challenges for AI safety work that relies on monitoring CoT to detect undesired behavior.

comment: Accepted to the Reasoning and Planning for Large Language Models Workshop (ICLR 25), 10 main paper pages, 38 appendix pages

♻ ★ Correlated Proxies: A New Definition and Improved Mitigation for Reward Hacking ICLR 2025

Cassidy Laidlaw, Shivam Singhal, Anca Dragan

Because it is difficult to precisely specify complex objectives, reinforcement learning policies are often optimized using proxy reward functions that only approximate the true goal. However, optimizing proxy rewards frequently leads to reward hacking: the optimized reward function ceases to be a good proxy and the resulting policy performs poorly with respect to the unspecified true reward. Principled solutions to reward hacking have been impeded by the lack of a good definition for the problem. To address this gap, we introduce a definition of reward hacking based on the correlation between proxy and true rewards for states and actions seen by a "reference policy" that breaks down under optimization. We show that this definition captures reward hacking behavior across several realistic settings, including in reinforcement learning from human feedback (RLHF). Using our formulation, we show theoretically that regularization to the reference policy can effectively prevent reward hacking. While the current practice in RLHF applies a KL penalty between action distributions for this purpose, our theory suggests regularizing the $\chi^2$ divergence between the policies' occupancy measures can be more effective. We intuitively show the benefits of this type of regularization and demonstrate that it better mitigates reward hacking in practice across four realistic settings, including RLHF. Our code is available at https://github.com/cassidylaidlaw/orpo.

comment: Spotlight at ICLR 2025

♻ ★ DataEnvGym: Data Generation Agents in Teacher Environments with Student Feedback ICLR 2025

Zaid Khan, Elias Stengel-Eskin, Jaemin Cho, Mohit Bansal

The process of creating training data to teach models is currently driven by humans, who manually analyze model weaknesses and plan how to create data that improves a student model. Approaches using LLMs as annotators reduce human effort, but still require humans to interpret feedback from evaluations and control the LLM to produce data the student needs. Automating this labor-intensive process by creating autonomous data generation agents - or teachers - is desirable, but requires environments that can simulate the feedback-driven, iterative, closed loop of data creation. To enable rapid, scalable testing for such agents and their modules, we introduce DataEnvGym, a testbed of teacher environments for data generation agents. DataEnvGym frames data generation as a sequential decision-making task, involving an agent consisting of a data generation policy (which generates a plan for creating training data) and a data generation engine (which transforms the plan into data), inside an environment that provides student feedback. The agent's goal is to improve student performance. Students are iteratively trained and evaluated on generated data, and their feedback (in the form of errors or weak skills) is reported to the agent after each iteration. DataEnvGym includes multiple teacher environment instantiations across 3 levels of structure in the state representation and action space. More structured environments are based on inferred skills and offer more interpretability and curriculum control. We support 4 domains (math, code, VQA, and tool-use) and test multiple students and teachers. Example agents in our teaching environments can iteratively improve students across tasks and settings. Moreover, we show that environments teach different skill levels and test variants of key modules, pointing to future work in improving data generation agents, engines, and feedback mechanisms.

comment: ICLR 2025 Spotlight; Project Page: https://DataEnvGym.github.io

♻ ★ What is the Alignment Objective of GRPO?

Milan Vojnovic, Se-Young Yun

In this note, we examine the aggregation of preferences achieved by the Group Policy Optimisation (GRPO) algorithm, a reinforcement learning method used to train advanced artificial intelligence models such as DeepSeek-R1-Zero and DeepSeekMath. The GRPO algorithm trains a policy using a reward preference model, which is computed by sampling a set of outputs for a given context, observing the corresponding rewards, and applying shift-and-scale normalisation to these reward values. Additionally, it incorporates a penalty function to discourage deviations from a reference policy. We present a framework that enables us to characterise the stationary policies of the GRPO algorithm. This analysis reveals that the aggregation of preferences differs fundamentally from standard logarithmic pooling, which is implemented by other approaches such as RLHF. The precise form of preference aggregation arises from the way the reward preference model is defined and from the penalty function, which we show to essentially correspond to the reverse Kullback-Leibler (KL) divergence between the aggregation policy and the reference policy. Interestingly, we demonstrate that for groups of size two, the reward preference model corresponds to pairwise comparison preferences, similar to those in other alignment methods based on pairwise comparison feedback. We provide explicit characterisations of the aggregate preference for binary questions, for groups of size two, and in the limit of large group size. This provides insights into the dependence of the aggregate preference on parameters such as the regularisation constant and the confidence margin of question answers. Finally, we discuss the aggregation of preferences obtained by modifying the GRPO algorithm to use direct KL divergence as the penalty or to use rewards without scale normalisation.

♻ ★ Representation Retrieval Learning for Heterogeneous Data Integration

Qi Xu, Annie Qu

In the era of big data, large-scale, multi-modal datasets are increasingly ubiquitous, offering unprecedented opportunities for predictive modeling and scientific discovery. However, these datasets often exhibit complex heterogeneity, such as covariate shift, posterior drift, and missing modalities, that can hinder the accuracy of existing prediction algorithms. To address these challenges, we propose a novel Representation Retrieval ($R^2$) framework, which integrates a representation learning module (the representer) with a sparsity-induced machine learning model (the learner). Moreover, we introduce the notion of "integrativeness" for representers, characterized by the effective data sources used in learning representers, and propose a Selective Integration Penalty (SIP) to explicitly improve the property. Theoretically, we demonstrate that the $R^2$ framework relaxes the conventional full-sharing assumption in multi-task learning, allowing for partially shared structures, and that SIP can improve the convergence rate of the excess risk bound. Extensive simulation studies validate the empirical performance of our framework, and applications to two real-world datasets further confirm its superiority over existing approaches.

♻ ★ Joint Fine-tuning and Conversion of Pretrained Speech and Language Models towards Linear Complexity ICLR 2025

Mutian He, Philip N. Garner

Architectures such as Linformer and Mamba have recently emerged as competitive linear time replacements for transformers. However, corresponding large pretrained models are often unavailable, especially in non-text domains. To remedy this, we present a Cross-Architecture Layerwise Distillation (CALD) approach that jointly converts a transformer model to a linear time substitute and fine-tunes it to a target task. We also compare several means to guide the fine-tuning to optimally retain the desired inference capability from the original model. The methods differ in their use of the target model and the trajectory of the parameters. In a series of empirical studies on language processing, language modeling, and speech processing, we show that CALD can effectively recover the result of the original model, and that the guiding strategy contributes to the result. Some reasons for the variation are suggested.

comment: 18 pages, 5 figures; ICLR 2025 camera ready. Code: https://github.com/idiap/linearize-distill-pretrained-transformers

♻ ★ Latent Space Chain-of-Embedding Enables Output-free LLM Self-Evaluation ICLR 2025

Yiming Wang, Pei Zhang, Baosong Yang, Derek F. Wong, Rui Wang

LLM self-evaluation relies on the LLM's own ability to estimate response correctness, which can greatly improve its deployment reliability. In this research track, we propose the Chain-of-Embedding (CoE) in the latent space to enable LLMs to perform output-free self-evaluation. CoE consists of all progressive hidden states produced during the inference time, which can be treated as the latent thinking path of LLMs. We find that when LLMs respond correctly and incorrectly, their CoE features differ, these discrepancies assist us in estimating LLM response correctness. Experiments in four diverse domains and seven LLMs fully demonstrate the effectiveness of our method. Meanwhile, its label-free design intent without any training and millisecond-level computational cost ensures real-time feedback in large-scale scenarios. More importantly, we provide interesting insights into LLM response correctness from the perspective of hidden state changes inside LLMs.

comment: Accepted by ICLR 2025

♻ ★ scMEDAL for the interpretable analysis of single-cell transcriptomics data with batch effect visualization using a deep mixed effects autoencoder

Aixa X. Andrade, Son Nguyen, Albert Montillo

scRNA-seq data has the potential to provide new insights into cellular heterogeneity and data acquisition; however, a major challenge is unraveling confounding from technical and biological batch effects. Existing batch correction algorithms suppress and discard these effects, rather than quantifying and modeling them. Here, we present scMEDAL, a framework for single-cell Mixed Effects Deep Autoencoder Learning, which separately models batch-invariant and batch-specific effects using two complementary autoencoder networks. One network is trained through adversarial learning to capture a batch-invariant representation, while a Bayesian autoencoder learns a batch-specific representation. Comprehensive evaluations spanning conditions (e.g., autism, leukemia, and cardiovascular), cell types, and technical and biological effects demonstrate that scMEDAL suppresses batch effects while modeling batch-specific variation, enhancing accuracy and interpretability. Unlike prior approaches, the framework's fixed- and random-effects autoencoders enable retrospective analyses, including predicting a cell's expression as if it had been acquired in a different batch via genomap projections at the cellular level, revealing the impact of biological (e.g., diagnosis) and technical (e.g., acquisition) effects. By combining scMEDAL's batch-agnostic and batch-specific latent spaces, it enables more accurate predictions of disease status, donor group, and cell type, making scMEDAL a valuable framework for gaining deeper insight into data acquisition and cellular heterogeneity.

comment: Main manuscript: 28 pages, including 8 figures and 1 table. Supplemental material: 19 pages

♻ ★ Confidence-Controlled Exploration: Efficient Sparse-Reward Policy Learning for Robot Navigation

Bhrij Patel, Kasun Weerakoon, Wesley A. Suttle, Alec Koppel, Brian M. Sadler, Tianyi Zhou, Amrit Singh Bedi, Dinesh Manocha

Reinforcement learning (RL) is a promising approach for robotic navigation, allowing robots to learn through trial and error. However, real-world robotic tasks often suffer from sparse rewards, leading to inefficient exploration and suboptimal policies due to sample inefficiency of RL. In this work, we introduce Confidence-Controlled Exploration (CCE), a novel method that improves sample efficiency in RL-based robotic navigation without modifying the reward function. Unlike existing approaches, such as entropy regularization and reward shaping, which can introduce instability by altering rewards, CCE dynamically adjusts trajectory length based on policy entropy. Specifically, it shortens trajectories when uncertainty is high to enhance exploration and extends them when confidence is high to prioritize exploitation. CCE is a principled and practical solution inspired by a theoretical connection between policy entropy and gradient estimation. It integrates seamlessly with on-policy and off-policy RL methods and requires minimal modifications. We validate CCE across REINFORCE, PPO, and SAC in both simulated and real-world navigation tasks. CCE outperforms fixed-trajectory and entropy-regularized baselines, achieving an 18\% higher success rate, 20-38\% shorter paths, and 9.32\% lower elevation costs under a fixed training sample budget. Finally, we deploy CCE on a Clearpath Husky robot, demonstrating its effectiveness in complex outdoor environments.

comment: 10 pages, 6 figures, 2 tables

♻ ★ Video Super-Resolution: All You Need is a Video Diffusion Model

Zhihao Zhan, Wang Pang, Xiang Zhu, Yechao Bai

We present a generic video super-resolution algorithm in this paper, based on the Diffusion Posterior Sampling framework with an unconditional video generation model in latent space. The video generation model, a diffusion transformer, functions as a space-time model. We argue that a powerful model, which learns the physics of the real world, can easily handle various kinds of motion patterns as prior knowledge, thus eliminating the need for explicit estimation of optical flows or motion parameters for pixel alignment. Furthermore, a single instance of the proposed video diffusion transformer model can adapt to different sampling conditions without re-training. Empirical results on synthetic and real-world datasets demonstrate that our method has strong capabilities to address video super-resolution challenges.

♻ ★ Fast MRI for All: Bridging Equity Gaps via Training without Raw Data Access

Yaşar Utku Alçalar, Merve Gülle, Mehmet Akçakaya

Physics-driven deep learning (PD-DL) approaches have become popular for improved reconstruction of fast magnetic resonance imaging (MRI) scans. Though PD-DL offers higher acceleration rates than existing clinical fast MRI techniques, their use has been limited outside specialized MRI centers. A key challenge is generalization to underrepresented pathologies or populations, noted in multiple studies, with fine-tuning on target populations suggested for improvement. However, current approaches for PD-DL training require access to raw k-space measurements, which is typically only available at specialized MRI centers that have research agreements for such data access. This is especially an issue for rural and underserved areas, where commercial MRI scanners only provide access to a final reconstructed image. To tackle these challenges, we propose Compressibility-inspired Unsupervised Learning via Parallel Imaging Fidelity (CUPID) for high-quality PD-DL training using only routine clinical reconstructed images exported from an MRI scanner. CUPID evaluates output quality with a compressibility-based approach while ensuring that the output stays consistent with the clinical parallel imaging reconstruction through well-designed perturbations. Our results show CUPID achieves similar quality to established PD-DL training that requires k-space data while outperforming compressed sensing (CS) and diffusion-based generative methods. We further demonstrate its effectiveness in a zero-shot training setup for retrospectively and prospectively sub-sampled acquisitions, attesting to its minimal training burden. As an approach that radically deviates from existing strategies, CUPID presents an opportunity to provide equitable access to fast MRI for underserved populations in an attempt to reduce the inequalities associated with this expensive imaging modality.

♻ ★ A Clifford Algebraic Approach to E(n)-Equivariant High-order Graph Neural Networks

Viet-Hoang Tran, Thieu N. Vo, Tho Tran Huu, Tan Minh Nguyen

Designing neural network architectures that can handle data symmetry is crucial. This is especially important for geometric graphs whose properties are equivariance under Euclidean transformations. Current equivariant graph neural networks (EGNNs), particularly those using message passing, have a limitation in expressive power. Recent high-order graph neural networks can overcome this limitation, yet they lack equivariance properties, representing a notable drawback in certain applications in chemistry and physical sciences. In this paper, we introduce the Clifford Group Equivariant Graph Neural Networks (CG-EGNNs), a novel EGNN that enhances high-order message passing by integrating high-order local structures in the context of Clifford algebras. As a key benefit of using Clifford algebras, CG-EGNN can learn functions that capture equivariance from positional features. By adopting the high-order message passing mechanism, CG-EGNN gains richer information from neighbors, thus improving model performance. Furthermore, we establish the universality property of the $k$-hop message passing framework, showcasing greater expressive power of CG-EGNNs with additional $k$-hop message passing mechanism. We empirically validate that CG-EGNNs outperform previous methods on various benchmarks including n-body, CMU motion capture, and MD17, highlighting their effectiveness in geometric deep learning.

♻ ★ Monomial Matrix Group Equivariant Neural Functional Networks NeurIPS 2024

Viet-Hoang Tran, Thieu N. Vo, Tho H. Tran, An T. Nguyen, Tan M. Nguyen

Neural functional networks (NFNs) have recently gained significant attention due to their diverse applications, ranging from predicting network generalization and network editing to classifying implicit neural representation. Previous NFN designs often depend on permutation symmetries in neural networks' weights, which traditionally arise from the unordered arrangement of neurons in hidden layers. However, these designs do not take into account the weight scaling symmetries of $\ReLU$ networks, and the weight sign flipping symmetries of $\sin$ or $\Tanh$ networks. In this paper, we extend the study of the group action on the network weights from the group of permutation matrices to the group of monomial matrices by incorporating scaling/sign-flipping symmetries. Particularly, we encode these scaling/sign-flipping symmetries by designing our corresponding equivariant and invariant layers. We name our new family of NFNs the Monomial Matrix Group Equivariant Neural Functional Networks (Monomial-NFN). Because of the expansion of the symmetries, Monomial-NFN has much fewer independent trainable parameters compared to the baseline NFNs in the literature, thus enhancing the model's efficiency. Moreover, for fully connected and convolutional neural networks, we theoretically prove that all groups that leave these networks invariant while acting on their weight spaces are some subgroups of the monomial matrix group. We provide empirical evidence to demonstrate the advantages of our model over existing baselines, achieving competitive performance and efficiency.

comment: 10 pages in the main text. Published at NeurIPS 2024. The code is available at https://github.com/MathematicalAI-NUS/Monomial-NFN

♻ ★ Class-wise Federated Unlearning: Harnessing Active Forgetting with Teacher-Student Memory Generation

Yuyuan Li, Jiaming Zhang, Yixiu Liu, Chaochao Chen

Privacy concerns associated with machine learning models have driven research into machine unlearning, which aims to erase the memory of specific target training data from already trained models. This issue also arises in federated learning, creating the need to address the federated unlearning problem. However, federated unlearning remains a challenging task. On the one hand, current research primarily focuses on unlearning all data from a client, overlooking more fine-grained unlearning targets, e.g., class-wise and sample-wise removal. On the other hand, existing methods suffer from imprecise estimation of data influence and impose significant computational or storage burden. To address these issues, we propose a neuro-inspired federated unlearning framework based on active forgetting, which is independent of model architectures and suitable for fine-grained unlearning targets. Our framework distinguishes itself from existing methods by utilizing new memories to overwrite old ones. These new memories are generated through teacher-student learning. We further utilize refined elastic weight consolidation to mitigate catastrophic forgetting of non-target data. Extensive experiments on benchmark datasets demonstrate the efficiency and effectiveness of our method, achieving satisfactory unlearning completeness against backdoor attacks.

♻ ★ Similarity Equivariant Graph Neural Networks for Homogenization of Metamaterials

Fleur Hendriks, Vlado Menkovski, Martin Doškář, Marc G. D. Geers, Ondřej Rokoš

Soft, porous mechanical metamaterials exhibit pattern transformations that may have important applications in soft robotics, sound reduction and biomedicine. To design these innovative materials, it is important to be able to simulate them accurately and quickly, in order to tune their mechanical properties. Since conventional simulations using the finite element method entail a high computational cost, in this article we aim to develop a machine learning-based approach that scales favorably to serve as a surrogate model. To ensure that the model is also able to handle various microstructures, including those not encountered during training, we include the microstructure as part of the network input. Therefore, we introduce a graph neural network that predicts global quantities (energy, stress stiffness) as well as the pattern transformations that occur (the kinematics). To make our model as accurate and data-efficient as possible, various symmetries are incorporated into the model. The starting point is an E(n)-equivariant graph neural network (which respects translation, rotation and reflection) that has periodic boundary conditions (i.e., it is in-/equivariant with respect to the choice of RVE), is scale in-/equivariant, can simulate large deformations, and can predict scalars, vectors as well as second and fourth order tensors (specifically energy, stress and stiffness). The incorporation of scale equivariance makes the model equivariant with respect to the similarities group, of which the Euclidean group E(n) is a subgroup. We show that this network is more accurate and data-efficient than graph neural networks with fewer symmetries. To create an efficient graph representation of the finite element discretization, we use only the internal geometrical hole boundaries from the finite element mesh to achieve a better speed-up and scaling with the mesh size.

comment: 60 pages, 22 figures. Published in CMAME (Computer Methods in Applied Mechanics and Engineering)

♻ ★ The Federation Strikes Back: A Survey of Federated Learning Privacy Attacks, Defenses, Applications, and Policy Landscape

Joshua C. Zhao, Saurabh Bagchi, Salman Avestimehr, Kevin S. Chan, Somali Chaterji, Dimitris Dimitriadis, Jiacheng Li, Ninghui Li, Arash Nourian, Holger R. Roth

Deep learning has shown incredible potential across a wide array of tasks, and accompanied by this growth has been an insatiable appetite for data. However, a large amount of data needed for enabling deep learning is stored on personal devices, and recent concerns on privacy have further highlighted challenges for accessing such data. As a result, federated learning (FL) has emerged as an important privacy-preserving technology that enables collaborative training of machine learning models without the need to send the raw, potentially sensitive, data to a central server. However, the fundamental premise that sending model updates to a server is privacy-preserving only holds if the updates cannot be "reverse engineered" to infer information about the private training data. It has been shown under a wide variety of settings that this privacy premise does not hold. In this survey paper, we provide a comprehensive literature review of the different privacy attacks and defense methods in FL. We identify the current limitations of these attacks and highlight the settings in which the privacy of ann FL client can be broken. We further dissect some of the successful industry applications of FL and draw lessons for future successful adoption. We survey the emerging landscape of privacy regulation for FL and conclude with future directions for taking FL toward the cherished goal of generating accurate models while preserving the privacy of the data from its participants.

comment: Accepted to ACM Computing Surveys; 35 pages

♻ ★ Uncertainty-Aware Robust Learning on Noisy Graphs ICASSP 2025

Shuyi Chen, Kaize Ding, Shixiang Zhu

Graph neural networks (GNNs) have excelled in various graph learning tasks, particularly node classification. However, their performance is often hampered by noisy measurements in real-world graphs, which can corrupt critical patterns in the data. To address this, we propose a novel uncertainty-aware graph learning framework inspired by distributionally robust optimization. Specifically, we use a graph neural network-based encoder to embed the node features and find the optimal node embeddings by minimizing the worst-case risk through a minimax formulation. Such an uncertainty-aware learning process leads to improved node representations and a more robust graph predictive model that effectively mitigates the impact of uncertainty arising from data noise. Our experimental results demonstrate superior predictive performance over baselines across noisy scenarios.

comment: ICASSP 2025 camera ready

♻ ★ Networked Communication for Decentralised Agents in Mean-Field Games

Patrick Benjamin, Alessandro Abate

We introduce networked communication to the mean-field game framework, in particular to oracle-free settings where $N$ decentralised agents learn along a single, non-episodic run of the empirical system. We prove that our architecture has sample guarantees bounded between those of the centralised- and independent-learning cases. We provide the order of the difference in these bounds in terms of network structure and number of communication rounds, and also contribute a policy-update stability guarantee. We discuss how the sample guarantees of the three theoretical algorithms do not actually result in practical convergence. We therefore show that in practical settings where the theoretical parameters are not observed (leading to poor estimation of the Q-function), our communication scheme considerably accelerates learning over the independent case, often performing similarly to a centralised learner while removing the restrictive assumption of the latter. We contribute further practical enhancements to all three theoretical algorithms, allowing us to present their first empirical demonstrations. Our experiments confirm that we can remove several of the theoretical assumptions of the algorithms, and display the empirical convergence benefits brought by our new networked communication. We additionally show that our networked approach has significant advantages over both alternatives in terms of robustness to update failures and to changes in population size.

♻ ★ Multilevel Generative Samplers for Investigating Critical Phenomena ICLR 2025

Ankur Singha, Elia Cellini, Kim A. Nicoli, Karl Jansen, Stefan Kühn, Shinichi Nakajima

Investigating critical phenomena or phase transitions is of high interest in physics and chemistry, for which Monte Carlo (MC) simulations, a crucial tool for numerically analyzing macroscopic properties of given systems, are often hindered by an emerging divergence of correlation length -- known as scale invariance at criticality (SIC) in the renormalization group theory. SIC causes the system to behave the same at any length scale, from which many existing sampling methods suffer: long-range correlations cause critical slowing down in Markov chain Monte Carlo (MCMC), and require intractably large receptive fields for generative samplers. In this paper, we propose a Renormalization-informed Generative Critical Sampler (RiGCS) -- a novel sampler specialized for near-critical systems, where SIC is leveraged as an advantage rather than a nuisance. Specifically, RiGCS builds on MultiLevel Monte Carlo (MLMC) with Heat Bath (HB) algorithms, which perform ancestral sampling from low-resolution to high-resolution lattice configurations with site-wise-independent conditional HB sampling. Although MLMC-HB is highly efficient under exact SIC, it suffers from a low acceptance rate under slight SIC violation. Notably, SIC violation always occurs in finite-size systems, and may induce long-range and higher-order interactions in the renormalized distributions, which are not considered by independent HB samplers. RiGCS enhances MLMC-HB by replacing a part of the conditional HB sampler with generative models that capture those residual interactions and improve the sampling efficiency. Our experiments show that the effective sample size of RiGCS is a few orders of magnitude higher than state-of-the-art generative model baselines in sampling configurations for 128x128 two-dimensional Ising systems.

comment: 10 pages, 4 figures (main text); 13th International Conference on Learning Representations (ICLR 2025)

♻ ★ Exploring a Multimodal Fusion-based Deep Learning Network for Detecting Facial Palsy IJCAI 2024

Heng Yim Nicole Oo, Min Hun Lee, Jeong Hoon Lim

Algorithmic detection of facial palsy offers the potential to improve current practices, which usually involve labor-intensive and subjective assessment by clinicians. In this paper, we present a multimodal fusion-based deep learning model that utilizes unstructured data (i.e. an image frame with facial line segments) and structured data (i.e. features of facial expressions) to detect facial palsy. We then contribute to a study to analyze the effect of different data modalities and the benefits of a multimodal fusion-based approach using videos of 21 facial palsy patients. Our experimental results show that among various data modalities (i.e. unstructured data - RGB images and images of facial line segments and structured data - coordinates of facial landmarks and features of facial expressions), the feed-forward neural network using features of facial expression achieved the highest precision of 76.22 while the ResNet-based model using images of facial line segments achieved the highest recall of 83.47. When we leveraged both images of facial line segments and features of facial expressions, our multimodal fusion-based deep learning model slightly improved the precision score to 77.05 at the expense of a decrease in the recall score.

comment: IJCAI 2024 4th AI for Ageless Aging Workshop (AIAA)

♻ ★ Unlocking Historical Clinical Trial Data with ALIGN: A Compositional Large Language Model System for Medical Coding

Nabeel Seedat, Caterina Tozzi, Andrea Hita Ardiaca, Mihaela van der Schaar, James Weatherall, Adam Taylor

The reuse of historical clinical trial data has significant potential to accelerate medical research and drug development. However, interoperability challenges, particularly with missing medical codes, hinders effective data integration across studies. While Large Language Models (LLMs) offer a promising solution for automated coding without labeled data, current approaches face challenges on complex coding tasks. We introduce ALIGN, a novel compositional LLM-based system for automated, zero-shot medical coding. ALIGN follows a three-step process: (1) diverse candidate code generation; (2) self-evaluation of codes and (3) confidence scoring and uncertainty estimation enabling human deferral to ensure reliability. We evaluate ALIGN on harmonizing medication terms into Anatomical Therapeutic Chemical (ATC) and medical history terms into Medical Dictionary for Regulatory Activities (MedDRA) codes extracted from 22 immunology trials. ALIGN outperformed the LLM baselines, while also providing capabilities for trustworthy deployment. For MedDRA coding, ALIGN achieved high accuracy across all levels, matching RAG and excelling at the most specific levels (87-90% for HLGT). For ATC coding, ALIGN demonstrated superior performance, particularly at lower hierarchy levels (ATC Level 4), with 72-73% overall accuracy and 86-89% accuracy for common medications, outperforming baselines by 7-22%. ALIGN's uncertainty-based deferral improved accuracy by 17% to 90% accuracy with 30% deferral, notably enhancing performance on uncommon medications. ALIGN achieves this cost-efficiently at \$0.0007 and \$0.02 per code for GPT-4o-mini and GPT-4o, reducing barriers to clinical adoption. ALIGN advances automated medical coding for clinical trial data, contributing to enhanced data interoperability and reusability, positioning it as a promising tool to improve clinical research and accelerate drug development.

♻ ★ Networked Communication for Mean-Field Games with Function Approximation and Empirical Mean-Field Estimation

Patrick Benjamin, Alessandro Abate

Recent algorithms allow decentralised agents, possibly connected via a communication network, to learn equilibria in Mean-Field Games from a non-episodic run of the empirical system. However, these algorithms are for tabular settings: this computationally limits the size of agents' observation space, meaning the algorithms cannot handle anything but small state spaces, nor generalise beyond policies depending only on the agent's local state to so-called 'population-dependent' policies. We address this limitation by introducing function approximation to the existing setting, drawing on the Munchausen Online Mirror Descent method that has previously been employed only in finite-horizon, episodic, centralised settings. While this permits us to include the mean field in the observation for players' policies, it is unrealistic to assume decentralised agents have access to this global information: we therefore also provide new algorithms allowing agents to locally estimate the global empirical distribution, and to improve this estimate via inter-agent communication. We show theoretically that exchanging policy information helps networked agents outperform both independent and even centralised agents in function-approximation settings. Our experiments demonstrate this happening empirically, by an even greater margin than in tabular settings, and show that the communication network allows decentralised agents to estimate the mean field for population-dependent policies.

♻ ★ Adaptive Split Learning over Energy-Constrained Wireless Edge Networks

Zuguang Li, Wen Wu, Shaohua Wu, Wei Wang

Split learning (SL) is a promising approach for training artificial intelligence (AI) models, in which devices collaborate with a server to train an AI model in a distributed manner, based on a same fixed split point. However, due to the device heterogeneity and variation of channel conditions, this way is not optimal in training delay and energy consumption. In this paper, we design an adaptive split learning (ASL) scheme which can dynamically select split points for devices and allocate computing resource for the server in wireless edge networks. We formulate an optimization problem to minimize the average training latency subject to long-term energy consumption constraint. The difficulties in solving this problem are the lack of future information and mixed integer programming (MIP). To solve it, we propose an online algorithm leveraging the Lyapunov theory, named OPEN, which decomposes it into a new MIP problem only with the current information. Then, a two-layer optimization method is proposed to solve the MIP problem. Extensive simulation results demonstrate that the ASL scheme can reduce the average training delay and energy consumption by 53.7% and 22.1%, respectively, as compared to the existing SL schemes.

comment: 6 pages, 5 figures, 20 conferences

♻ ★ Respecting the limit:Bayesian optimization with a bound on the optimal value

Hanyang Wang, Juergen Branke, Matthias Poloczek

In many real-world optimization problems, we have prior information about what objective function values are achievable. In this paper, we study the scenario that we have either exact knowledge of the minimum value or a, possibly inexact, lower bound on its value. We propose bound-aware Bayesian optimization (BABO), a Bayesian optimization method that uses a new surrogate model and acquisition function to utilize such prior information. We present SlogGP, a new surrogate model that incorporates bound information and adapts the Expected Improvement (EI) acquisition function accordingly. Empirical results on a variety of benchmarks demonstrate the benefit of taking prior information about the optimal value into account, and that the proposed approach significantly outperforms existing techniques. Furthermore, we notice that even in the absence of prior information on the bound, the proposed SlogGP surrogate model still performs better than the standard GP model in most cases, which we explain by its larger expressiveness.

♻ ★ Neuroplastic Expansion in Deep Reinforcement Learning

Jiashun Liu, Johan Obando-Ceron, Aaron Courville, Ling Pan

The loss of plasticity in learning agents, analogous to the solidification of neural pathways in biological brains, significantly impedes learning and adaptation in reinforcement learning due to its non-stationary nature. To address this fundamental challenge, we propose a novel approach, {\it Neuroplastic Expansion} (NE), inspired by cortical expansion in cognitive science. NE maintains learnability and adaptability throughout the entire training process by dynamically growing the network from a smaller initial size to its full dimension. Our method is designed with three key components: (\textit{1}) elastic topology generation based on potential gradients, (\textit{2}) dormant neuron pruning to optimize network expressivity, and (\textit{3}) neuron consolidation via experience review to strike a balance in the plasticity-stability dilemma. Extensive experiments demonstrate that NE effectively mitigates plasticity loss and outperforms state-of-the-art methods across various tasks in MuJoCo and DeepMind Control Suite environments. NE enables more adaptive learning in complex, dynamic environments, which represents a crucial step towards transitioning deep reinforcement learning from static, one-time training paradigms to more flexible, continually adapting models.

♻ ★ Dark Deceptions in DHCP: Dismantling Network Defenses

Robert Dilworth

This paper explores vulnerabilities in the Dynamic Host Configuration Protocol (DHCP) and their implications on the Confidentiality, Integrity, and Availability (CIA) Triad. Through an analysis of various attacks, including DHCP Starvation, Rogue DHCP Servers, Replay Attacks, and TunnelVision exploits, the paper provides a taxonomic classification of threats, assesses risks, and proposes appropriate controls. The discussion also highlights the dangers of VPN decloaking through DHCP exploits and underscores the importance of safeguarding network infrastructures. By bringing awareness to the TunnelVision exploit, this paper aims to mitigate risks associated with these prevalent vulnerabilities.

comment: 8 pages, 4 tables

♻ ★ Diabetica: Adapting Large Language Model to Enhance Multiple Medical Tasks in Diabetes Care and Management ICLR 2025

Lai Wei, Zhen Ying, Muyang He, Yutong Chen, Qian Yang, Yanzhe Hong, Jiaping Lu, Kaipeng Zheng, Shaoting Zhang, Xiaoying Li, Weiran Huang, Ying Chen

Diabetes is a chronic disease with a significant global health burden, requiring multi-stakeholder collaboration for optimal management. Large language models (LLMs) have shown promise in various healthcare scenarios, but their effectiveness across diverse diabetes tasks remains unproven. Our study introduced a framework to train and validate diabetes-specific LLMs. We first developed a comprehensive data processing pipeline that includes data collection, filtering, augmentation and refinement. This created a high-quality, diabetes-specific dataset and evaluation benchmarks from scratch. Fine-tuned on the collected training dataset, our diabetes-specific LLM family demonstrated state-of-the-art proficiency in processing various diabetes tasks compared to other LLMs. Furthermore, clinical studies revealed the potential applications of our models in diabetes care, including providing personalized healthcare, assisting medical education, and streamlining clinical tasks. Generally, our introduced framework helps develop diabetes-specific LLMs and highlights their potential to enhance clinical practice and provide personalized, data-driven support for diabetes management across different end users. Our codes, benchmarks and models are available at https://github.com/waltonfuture/Diabetica.

comment: Accepted by ICLR 2025 SCI-FM workshop

♻ ★ Towards Class-wise Robustness Analysis

Tejaswini Medi, Julia Grabinski, Margret Keuper

While being very successful in solving many downstream tasks, the application of deep neural networks is limited in real-life scenarios because of their susceptibility to domain shifts such as common corruptions, and adversarial attacks. The existence of adversarial examples and data corruption significantly reduces the performance of deep classification models. Researchers have made strides in developing robust neural architectures to bolster decisions of deep classifiers. However, most of these works rely on effective adversarial training methods, and predominantly focus on overall model robustness, disregarding class-wise differences in robustness, which are critical. Exploiting weakly robust classes is a potential avenue for attackers to fool the image recognition models. Therefore, this study investigates class-to-class biases across adversarially trained robust classification models to understand their latent space structures and analyze their strong and weak class-wise properties. We further assess the robustness of classes against common corruptions and adversarial attacks, recognizing that class vulnerability extends beyond the number of correct classifications for a specific class. We find that the number of false positives of classes as specific target classes significantly impacts their vulnerability to attacks. Through our analysis on the Class False Positive Score, we assess a fair evaluation of how susceptible each class is to misclassification.

♻ ★ Deep Reinforcement Learning for Dynamic Resource Allocation in Wireless Networks

Shubham Malhotra, Fnu Yashu, Muhammad Saqib, Dipkumar Mehta, Jagdish Jangid, Sachin Dixit

This report investigates the application of deep reinforcement learning (DRL) algorithms for dynamic resource allocation in wireless communication systems. An environment that includes a base station, multiple antennas, and user equipment is created. Using the RLlib library, various DRL algorithms such as Deep Q-Network (DQN) and Proximal Policy Optimization (PPO) are then applied. These algorithms are compared based on their ability to optimize resource allocation, focusing on the impact of different learning rates and scheduling policies. The findings demonstrate that the choice of algorithm and learning rate significantly influences system performance, with DRL providing more efficient resource allocation compared to traditional methods.

comment: Upon further review, we found inconsistencies in our analysis and decided to conduct additional research before resubmitting a revised version

♻ ★ Automated Knowledge Concept Annotation and Question Representation Learning for Knowledge Tracing

Yilmazcan Ozyurt, Stefan Feuerriegel, Mrinmaya Sachan

Knowledge tracing (KT) is a popular approach for modeling students' learning progress over time, which can enable more personalized and adaptive learning. However, existing KT approaches face two major limitations: (1) they rely heavily on expert-defined knowledge concepts (KCs) in questions, which is time-consuming and prone to errors; and (2) KT methods tend to overlook the semantics of both questions and the given KCs. In this work, we address these challenges and present KCQRL, a framework for automated knowledge concept annotation and question representation learning that can improve the effectiveness of any existing KT model. First, we propose an automated KC annotation process using large language models (LLMs), which generates question solutions and then annotates KCs in each solution step of the questions. Second, we introduce a contrastive learning approach to generate semantically rich embeddings for questions and solution steps, aligning them with their associated KCs via a tailored false negative elimination approach. These embeddings can be readily integrated into existing KT models, replacing their randomly initialized embeddings. We demonstrate the effectiveness of KCQRL across 15 KT algorithms on two large real-world Math learning datasets, where we achieve consistent performance improvements.

♻ ★ Low-Rank Continual Personalization of Diffusion Models ICLR 2025

Łukasz Staniszewski, Katarzyna Zaleska, Kamil Deja

Recent personalization methods for diffusion models, such as Dreambooth and LoRA, allow fine-tuning pre-trained models to generate new concepts. However, applying these techniques across consecutive tasks in order to include, e.g., new objects or styles, leads to a forgetting of previous knowledge due to mutual interference between their adapters. In this work, we tackle the problem of continual customization under a rigorous regime with no access to past tasks' adapters. In such a scenario, we investigate how different adapters' initialization and merging methods can improve the quality of the final model. To that end, we evaluate the naive continual fine-tuning of customized models and compare this approach with three methods for consecutive adapters' training: sequentially merging new adapters, merging orthogonally initialized adapters, and updating only relevant task-specific weights. In our experiments, we show that the proposed techniques mitigate forgetting when compared to the naive approach. In our studies, we show different traits of selected techniques and their effect on the plasticity and stability of the continually adapted model. Repository with the code is available at https://github.com/luk-st/continual-lora.

comment: SCOPE @ ICLR 2025

♻ ★ A Triple-Inertial Accelerated Alternating Optimization Method for Deep Learning Training

Chengcheng Yan, Jiawei Xu, Qingsong Wang, Zheng Peng

The stochastic gradient descent (SGD) algorithm has achieved remarkable success in training deep learning models. However, it has several limitations, including susceptibility to vanishing gradients, sensitivity to input data, and a lack of robust theoretical guarantees. In recent years, alternating minimization (AM) methods have emerged as a promising alternative for model training by employing gradient-free approaches to iteratively update model parameters. Despite their potential, these methods often exhibit slow convergence rates. To address this challenge, we propose a novel Triple-Inertial Accelerated Alternating Minimization (TIAM) framework for neural network training. The TIAM approach incorporates a triple-inertial acceleration strategy with a specialized approximation method, facilitating targeted acceleration of different terms in each sub-problem optimization. This integration improves the efficiency of convergence, achieving superior performance with fewer iterations. Additionally, we provide a convergence analysis of the TIAM algorithm, including its global convergence properties and convergence rate. Extensive experiments validate the effectiveness of the TIAM method, showing significant improvements in generalization capability and computational efficiency compared to existing approaches, particularly when applied to the rectified linear unit (ReLU) and its variants.

♻ ★ AI-Aided Kalman Filters

Nir Shlezinger, Guy Revach, Anubhab Ghosh, Saikat Chatterjee, Shuo Tang, Tales Imbiriba, Jindrich Dunik, Ondrej Straka, Pau Closas, Yonina C. Eldar

The Kalman filter (KF) and its variants are among the most celebrated algorithms in signal processing. These methods are used for state estimation of dynamic systems by relying on mathematical representations in the form of simple state-space (SS) models, which may be crude and inaccurate descriptions of the underlying dynamics. Emerging data-centric artificial intelligence (AI) techniques tackle these tasks using deep neural networks (DNNs), which are model-agnostic. Recent developments illustrate the possibility of fusing DNNs with classic Kalman-type filtering, obtaining systems that learn to track in partially known dynamics. This article provides a tutorial-style overview of design approaches for incorporating AI in aiding KF-type algorithms. We review both generic and dedicated DNN architectures suitable for state estimation, and provide a systematic presentation of techniques for fusing AI tools with KFs and for leveraging partial SS modeling and data, categorizing design approaches into task-oriented and SS model-oriented. The usefulness of each approach in preserving the individual strengths of model-based KFs and data-driven DNNs is investigated in a qualitative and quantitative study, whose code is publicly available, illustrating the gains of hybrid model-based/data-driven designs. We also discuss existing challenges and future research directions that arise from fusing AI and Kalman-type algorithms.

comment: Submitted to the IEEE Signal Processing Magazine

♻ ★ Tackling water table depth modeling via machine learning: From proxy observations to verifiability

Joseph Janssen, Ardalan Tootchi, Ali A. Ameli

Spatial patterns of water table depth (WTD) play a crucial role in shaping ecological resilience, hydrological connectivity, and human-centric systems. Generally, a large-scale (e.g., continental or global) continuous map of static WTD can be simulated using either physically-based (PB) or machine learning-based (ML) models. We construct three fine-resolution (500 m) ML simulations of WTD, using the XGBoost algorithm and more than 20 million real and proxy observations of WTD, across the United States and Canada. The three ML models were constrained using known physical relations between WTD's drivers and WTD and were trained by sequentially adding real and proxy observations of WTD. Through an extensive (pixel-by-pixel) evaluation across the study region and within ten major ecoregions of North America, we demonstrate that our models (corr=0.6-0.75) can more accurately predict unseen real and proxy observations of WTD compared to two available PB simulations of WTD (corr=0.21-0.40). However, we still argue that currently-available large-scale simulations of static WTD could be uncertain within data-scarce regions such as steep mountainous regions. We reason that biased observational data mainly collected from low-elevation floodplains and the over-flexibility of available models can negatively affect the verifiability of large-scale simulations of WTD. Ultimately, we thoroughly discuss future directions that may help hydrogeologists decide how to improve machine learning-based WTD estimations. In particular, we advocate for the use of proxy satellite data, the incorporation of physical laws, the implementation of better model verification standards, the development of novel globally-available emergent indices, and the collection of more reliable observations.

♻ ★ RedChronos: A Large Language Model-Based Log Analysis System for Insider Threat Detection in Enterprises

Chenyu Li, Zhengjia Zhu, Jiyan He, Xiu Zhang

Internal threat detection (IDT) aims to address security threats within organizations or enterprises by identifying potential or already occurring malicious threats within vast amounts of logs. Although organizations or enterprises have dedicated personnel responsible for reviewing these logs, it is impossible to manually examine all logs entirely.In response to the vast number of logs, we propose a system called RedChronos, which is a Large Language Model-Based Log Analysis System. This system incorporates innovative improvements over previous research by employing Query-Aware Weighted Voting and a Semantic Expansion-based Genetic Algorithm with LLM-driven Mutations. On the public datasets CERT 4.2 and 5.2, RedChronos outperforms or matches existing approaches in terms of accuracy, precision, and detection rate. Moreover, RedChronos reduces the need for manual intervention in security log reviews by approximately 90% in the Xiaohongshu Security Operation Center. Therefore, our RedChronos system demonstrates exceptional performance in handling IDT tasks, providing innovative solutions for these challenges. We believe that future research can continue to enhance the system's performance in IDT tasks while also reducing the response time to internal risk events.

♻ ★ Histogram Approaches for Imbalanced Data Streams Regression

Ehsan Aminian, Rita P. Ribeiro, Joao Gama

Imbalanced domains pose a significant challenge in real-world predictive analytics, particularly in the context of regression. While existing research has primarily focused on batch learning from static datasets, limited attention has been given to imbalanced regression in online learning scenarios. Intending to address this gap, in prior work, we proposed sampling strategies based on Chebyshevs inequality as the first methodologies designed explicitly for data streams. However, these approaches operated under the restrictive assumption that rare instances exclusively reside at distribution extremes. This study introduces histogram-based sampling strategies to overcome this constraint, proposing flexible solutions for imbalanced regression in evolving data streams. The proposed techniques -- Histogram-based Undersampling (HistUS) and Histogram-based Oversampling (HistOS) -- employ incremental online histograms to dynamically detect and prioritize rare instances across arbitrary regions of the target distribution to improve predictions in the rare cases. Comprehensive experiments on synthetic and real-world benchmarks demonstrate that HistUS and HistOS substantially improve rare-case prediction accuracy, outperforming baseline models while maintaining competitiveness with Chebyshev-based approaches.

♻ ★ Credal Two-Sample Tests of Epistemic Uncertainty

Siu Lun Chau, Antonin Schrab, Arthur Gretton, Dino Sejdinovic, Krikamol Muandet

We introduce credal two-sample testing, a new hypothesis testing framework for comparing credal sets -- convex sets of probability measures where each element captures aleatoric uncertainty and the set itself represents epistemic uncertainty that arises from the modeller's partial ignorance. Compared to classical two-sample tests, which focus on comparing precise distributions, the proposed framework provides a broader and more versatile set of hypotheses. This approach enables the direct integration of epistemic uncertainty, effectively addressing the challenges arising from partial ignorance in hypothesis testing. By generalising two-sample test to compare credal sets, our framework enables reasoning for equality, inclusion, intersection, and mutual exclusivity, each offering unique insights into the modeller's epistemic beliefs. As the first work on nonparametric hypothesis testing for comparing credal sets, we focus on finitely generated credal sets derived from i.i.d. samples from multiple distributions -- referred to as credal samples. We formalise these tests as two-sample tests with nuisance parameters and introduce the first permutation-based solution for this class of problems, significantly improving existing methods. Our approach properly incorporates the modeller's epistemic uncertainty into hypothesis testing, leading to more robust and credible conclusions, with kernel-based implementations for real-world applications.

comment: 64 pages

♻ ★ Column-wise Quantization of Weights and Partial Sums for Accurate and Efficient Compute-In-Memory Accelerators

Jiyoon Kim, Kang Eun Jeon, Yulhwa Kim, Jong Hwan Ko

Compute-in-memory (CIM) is an efficient method for implementing deep neural networks (DNNs) but suffers from substantial overhead from analog-to-digital converters (ADCs), especially as ADC precision increases. Low-precision ADCs can reduce this overhead but introduce partial-sum quantization errors degrading accuracy. Additionally, low-bit weight constraints, imposed by cell limitations and the need for multiple cells for higher-bit weights, present further challenges. While fine-grained partial-sum quantization has been studied to lower ADC resolution effectively, weight granularity, which limits overall partial-sum quantized accuracy, remains underexplored. This work addresses these challenges by aligning weight and partial-sum quantization granularities at the column-wise level. Our method improves accuracy while maintaining dequantization overhead, simplifies training by removing two-stage processes, and ensures robustness to memory cell variations via independent column-wise scale factors. We also propose an open-source CIM-oriented convolution framework to handle fine-grained weights and partial-sums efficiently, incorporating a novel tiling method and group convolution. Experimental results on ResNet-20 (CIFAR-10, CIFAR-100) and ResNet-18 (ImageNet) show accuracy improvements of 0.99%, 2.69%, and 1.01%, respectively, compared to the best-performing related works. Additionally, variation analysis reveals the robustness of our method against memory cell variations. These findings highlight the effectiveness of our quantization scheme in enhancing accuracy and robustness while maintaining hardware efficiency in CIM-based DNN implementations. Our code is available at https://github.com/jiyoonkm/ColumnQuant.

♻ ★ Bayesian Experimental Design via Contrastive Diffusions

Jacopo Iollo, Christophe Heinkelé, Pierre Alliez, Florence Forbes

Bayesian Optimal Experimental Design (BOED) is a powerful tool to reduce the cost of running a sequence of experiments. When based on the Expected Information Gain (EIG), design optimization corresponds to the maximization of some intractable expected contrast between prior and posterior distributions. Scaling this maximization to high dimensional and complex settings has been an issue due to BOED inherent computational complexity. In this work, we introduce a pooled posterior distribution with cost-effective sampling properties and provide a tractable access to the EIG contrast maximization via a new EIG gradient expression. Diffusion-based samplers are used to compute the dynamics of the pooled posterior and ideas from bi-level optimization are leveraged to derive an efficient joint sampling-optimization loop. The resulting efficiency gain allows to extend BOED to the well-tested generative capabilities of diffusion models. By incorporating generative models into the BOED framework, we expand its scope and its use in scenarios that were previously impractical. Numerical experiments and comparison with state-of-the-art methods show the potential of the approach.

♻ ★ FlashRNN: I/O-Aware Optimization of Traditional RNNs on modern hardware

Korbinian Pöppel, Maximilian Beck, Sepp Hochreiter

While Transformers and other sequence-parallelizable neural network architectures seem like the current state of the art in sequence modeling, they specifically lack state-tracking capabilities. These are important for time-series tasks and logical reasoning. Traditional RNNs like LSTMs and GRUs, as well as modern variants like sLSTM do have these capabilities at the cost of strictly sequential processing. While this is often seen as a strong limitation, we show how fast these networks can get with our hardware-optimization FlashRNN in Triton and CUDA, optimizing kernels to the register level on modern GPUs. We extend traditional RNNs with a parallelization variant that processes multiple RNNs of smaller hidden state in parallel, similar to the head-wise processing in Transformers. To enable flexibility on different GPU variants, we introduce a new optimization framework for hardware-internal cache sizes, memory and compute handling. It models the hardware in a setting using polyhedral-like constraints, including the notion of divisibility. This speeds up the solution process in our ConstrINT library for general integer constraint satisfaction problems (integer CSPs). We show that our kernels can achieve 50x speed-ups over a vanilla PyTorch implementation and allow 40x larger hidden sizes compared to our Triton implementation. Our open-source kernels and the optimization library are released here to boost research in the direction of state-tracking enabled RNNs and sequence modeling: https://github.com/NX-AI/flashrnn

♻ ★ Hidden in the Noise: Two-Stage Robust Watermarking for Images

Kasra Arabi, Benjamin Feuer, R. Teal Witter, Chinmay Hegde, Niv Cohen

As the quality of image generators continues to improve, deepfakes become a topic of considerable societal debate. Image watermarking allows responsible model owners to detect and label their AI-generated content, which can mitigate the harm. Yet, current state-of-the-art methods in image watermarking remain vulnerable to forgery and removal attacks. This vulnerability occurs in part because watermarks distort the distribution of generated images, unintentionally revealing information about the watermarking techniques. In this work, we first demonstrate a distortion-free watermarking method for images, based on a diffusion model's initial noise. However, detecting the watermark requires comparing the initial noise reconstructed for an image to all previously used initial noises. To mitigate these issues, we propose a two-stage watermarking framework for efficient detection. During generation, we augment the initial noise with generated Fourier patterns to embed information about the group of initial noises we used. For detection, we (i) retrieve the relevant group of noises, and (ii) search within the given group for an initial noise that might match our image. This watermarking approach achieves state-of-the-art robustness to forgery and removal against a large battery of attacks.

♻ ★ Clipping Improves Adam-Norm and AdaGrad-Norm when the Noise Is Heavy-Tailed

Savelii Chezhegov, Yaroslav Klyukin, Andrei Semenov, Aleksandr Beznosikov, Alexander Gasnikov, Samuel Horváth, Martin Takáč, Eduard Gorbunov

Methods with adaptive stepsizes, such as AdaGrad and Adam, are essential for training modern Deep Learning models, especially Large Language Models. Typically, the noise in the stochastic gradients is heavy-tailed for the later ones. Gradient clipping provably helps to achieve good high-probability convergence for such noises. However, despite the similarity between AdaGrad/Adam and Clip-SGD, the current understanding of the high-probability convergence of AdaGrad/Adam-type methods is limited in this case. In this work, we prove that AdaGrad/Adam (and their delayed version) can have provably bad high-probability convergence if the noise is heavy-tailed. We also show that gradient clipping fixes this issue, i.e., we derive new high-probability convergence bounds with polylogarithmic dependence on the confidence level for AdaGrad-Norm and Adam-Norm with clipping and with/without delay for smooth convex/non-convex stochastic optimization with heavy-tailed noise. Our empirical evaluations highlight the superiority of clipped versions of AdaGrad/Adam-Norm in handling the heavy-tailed noise.

comment: 63 pages, 8 figures

♻ ★ Real-Time Recurrent Reinforcement Learning

Julian Lemmel, Radu Grosu

We introduce a biologically plausible RL framework for solving tasks in partially observable Markov decision processes (POMDPs). The proposed algorithm combines three integral parts: (1) A Meta-RL architecture, resembling the mammalian basal ganglia; (2) A biologically plausible reinforcement learning algorithm, exploiting temporal difference learning and eligibility traces to train the policy and the value-function; (3) An online automatic differentiation algorithm for computing the gradients with respect to parameters of a shared recurrent network backbone. Our experimental results show that the method is capable of solving a diverse set of partially observable reinforcement learning tasks. The algorithm we call real-time recurrent reinforcement learning (RTRRL) serves as a model of learning in biological neural networks, mimicking reward pathways in the basal ganglia.

comment: 14 pages, 9 figures, includes Appendix

♻ ★ Materials Map Integrating Experimental and Computational Data through Graph-Based Machine Learning for Enhanced Materials Discovery

Yusuke Hashimoto, Xue Jia, Hao Li, Takaaki Tomai

Materials informatics (MI), which emerges from the integration of materials science and data science, is expected to greatly streamline the material discovery and development. The data used for MI are obtained from both computational and experimental studies, while their integration remains challenging. In our previous study, we reported the integration of these datasets by applying a machine learning model that captures trends hidden in the experimental datasets to compositional data stored in the computational database. In this study, we use the obtained data to construct materials maps, which visualize the relation in the structural features of materials, aiming to support study by the experimental researchers. The map is constructed using the MatDeepLearn (MDL) framework, which implements the graph-based representation of material structures, deep learning, and dimensional reduction for the map construction. We evaluate the obtained materials maps through statistical analysis and found that the MDL using message passing neural network (MPNN) enables efficient extraction of features that reflect the structural complexity of materials. Moreover, we found that this advantage does not necessarily translate into improved accuracy in predicting material properties. We attribute this unexpected outcome to the high learning performance inherent in MPNN, which can contribute to the structuring of data points within the materials map.

♻ ★ Feasible Policy Iteration for Safe Reinforcement Learning

Yujie Yang, Zhilong Zheng, Shengbo Eben Li, Wei Xu, Jingjing Liu, Xianyuan Zhan, Ya-Qin Zhang

Safety is the priority concern when applying reinforcement learning (RL) algorithms to real-world control problems. While policy iteration provides a fundamental algorithm for standard RL, an analogous theoretical algorithm for safe RL remains absent. In this paper, we propose feasible policy iteration (FPI), the first foundational dynamic programming algorithm for safe RL. FPI alternates between policy evaluation, region identification and policy improvement. This follows actor-critic-scenery (ACS) framework where scenery refers to a feasibility function that represents a feasible region. A region-wise update rule is developed for the policy improvement step, which maximizes state-value function inside the feasible region and minimizes feasibility function outside it. With this update rule, FPI guarantees monotonic expansion of feasible region, monotonic improvement of state-value function, and geometric convergence to the optimal safe policy. Experimental results demonstrate that FPI achieves strictly zero constraint violation on low-dimensional tasks and outperforms existing methods in constraint adherence and reward performance on high-dimensional tasks.

♻ ★ The R2D2 Deep Neural Network Series for Scalable Non-Cartesian Magnetic Resonance Imaging

Yiwei Chen, Amir Aghabiglou, Shijie Chen, Motahare Torki, Chao Tang, Ruud B. van Heeswijk, Yves Wiaux

We introduce the R2D2 Deep Neural Network (DNN) series paradigm for fast and scalable image reconstruction from highly-accelerated non-Cartesian k-space acquisitions in Magnetic Resonance Imaging (MRI). While unrolled DNN architectures provide a robust image formation approach via data-consistency layers, embedding non-uniform fast Fourier transform operators in a DNN can become impractical to train at large scale, e.g in 2D MRI with a large number of coils, or for higher-dimensional imaging. Plug-and-play approaches that alternate a learned denoiser blind to the measurement setting with a data-consistency step are not affected by this limitation but their highly iterative nature implies slow reconstruction. To address this scalability challenge, we leverage the R2D2 paradigm that was recently introduced to enable ultra-fast reconstruction for large-scale Fourier imaging in radio astronomy. R2D2's reconstruction is formed as a series of residual images iteratively estimated as outputs of DNN modules taking the previous iteration's data residual as input. The method can be interpreted as a learned version of the Matching Pursuit algorithm. A series of R2D2 DNN modules were sequentially trained in a supervised manner on the fastMRI dataset and validated for 2D multi-coil MRI in simulation and on real data, targeting highly under-sampled radial k-space sampling. Results suggest that a series with only few DNNs achieves superior reconstruction quality over its unrolled incarnation R2D2-Net (whose training is also much less scalable), and over the state-of-the-art diffusion-based "Decomposed Diffusion Sampler" approach (also characterised by a slower reconstruction process).

comment: 13 pages, 10 figures

♻ ★ MarS: a Financial Market Simulation Engine Powered by Generative Foundation Model ICLR 2025

Junjie Li, Yang Liu, Weiqing Liu, Shikai Fang, Lewen Wang, Chang Xu, Jiang Bian

Generative models aim to simulate realistic effects of various actions across different contexts, from text generation to visual effects. Despite significant efforts to build real-world simulators, the application of generative models to virtual worlds, like financial markets, remains under-explored. In financial markets, generative models can simulate complex market effects of participants with various behaviors, enabling interaction under different market conditions, and training strategies without financial risk. This simulation relies on the finest structured data in financial market like orders thus building the finest realistic simulation. We propose Large Market Model (LMM), an order-level generative foundation model, for financial market simulation, akin to language modeling in the digital world. Our financial Market Simulation engine (MarS), powered by LMM, addresses the domain-specific need for realistic, interactive and controllable order generation. Key observations include LMM's strong scalability across data size and model complexity, and MarS's robust and practicable realism in controlled generation with market impact. We showcase MarS as a forecast tool, detection system, analysis platform, and agent training environment, thus demonstrating MarS's "paradigm shift" potential for a variety of financial applications. We release the code of MarS at https://github.com/microsoft/MarS/.

comment: 35 pages, 26 figures, ICLR 2025

♻ ★ Score matching for bridges without learning time-reversals

Elizabeth L. Baker, Moritz Schauer, Stefan Sommer

We propose a new algorithm for learning bridged diffusion processes using score-matching methods. Our method relies on reversing the dynamics of the forward process and using this to learn a score function, which, via Doob's $h$-transform, yields a bridged diffusion process; that is, a process conditioned on an endpoint. In contrast to prior methods, we learn the score term $\nabla_x \log p(t, x; T, y)$ directly, for given $t, y$, completely avoiding first learning a time-reversal. We compare the performance of our algorithm with existing methods and see that it outperforms using the (learned) time-reversals to learn the score term. The code can be found at https://github.com/libbylbaker/forward_bridge.

♻ ★ Causal Representation Learning from Multimodal Biomedical Observations

Yuewen Sun, Lingjing Kong, Guangyi Chen, Loka Li, Gongxu Luo, Zijian Li, Yixuan Zhang, Yujia Zheng, Mengyue Yang, Petar Stojanov, Eran Segal, Eric P. Xing, Kun Zhang

Prevalent in biomedical applications (e.g., human phenotype research), multimodal datasets can provide valuable insights into the underlying physiological mechanisms. However, current machine learning (ML) models designed to analyze these datasets often lack interpretability and identifiability guarantees, which are essential for biomedical research. Recent advances in causal representation learning have shown promise in identifying interpretable latent causal variables with formal theoretical guarantees. Unfortunately, most current work on multimodal distributions either relies on restrictive parametric assumptions or yields only coarse identification results, limiting their applicability to biomedical research that favors a detailed understanding of the mechanisms. In this work, we aim to develop flexible identification conditions for multimodal data and principled methods to facilitate the understanding of biomedical datasets. Theoretically, we consider a nonparametric latent distribution (c.f., parametric assumptions in previous work) that allows for causal relationships across potentially different modalities. We establish identifiability guarantees for each latent component, extending the subspace identification results from previous work. Our key theoretical contribution is the structural sparsity of causal connections between modalities, which, as we will discuss, is natural for a large collection of biomedical systems. Empirically, we present a practical framework to instantiate our theoretical insights. We demonstrate the effectiveness of our approach through extensive experiments on both numerical and synthetic datasets. Results on a real-world human phenotype dataset are consistent with established biomedical research, validating our theoretical and methodological framework.

♻ ★ Symmetries, Scaling Laws and Phase Transitions in Consumer Advertising Response

Javier Marin

Understanding how consumers respond to business advertising efforts is essential for optimizing marketing investment. This research introduces a new modeling approach based on the concepts of symmetries and scaling laws in physics to describe consumer response to advertising dynamics. Drawing from mathematical frameworks used in physics and social sciences, we propose a model that accounts for a key aspect: the saturation effect. The model is validated against commonly used models, including the Michaelis-Menten and Hill equations, showing its ability to better capture nonlinearities in advertising effects. We introduce new key parameters like Marketing Sensitivity, Response Sensitivity, and Behavioral Sensitivit, that offer additional insights into the drivers of audience engagement and advertising performance. Our model provides a rigorous yet practical tool for understanding audience behavior, contributing to the improvement of budget allocation strategies.

♻ ★ Jailbreaking Large Language Models in Infinitely Many Ways

Oliver Goldstein, Emanuele La Malfa, Felix Drinkall, Samuele Marro, Michael Wooldridge

We discuss the ``Infinitely Many Paraphrases'' attacks (IMP), a category of jailbreaks that leverages the increasing capabilities of a model to handle paraphrases and encoded communications to bypass their defensive mechanisms. IMPs' viability pairs and grows with a model's capabilities to handle and bind the semantics of simple mappings between tokens and work extremely well in practice, posing a concrete threat to the users of the most powerful LLMs in commerce. We show how one can bypass the safeguards of the most powerful open- and closed-source LLMs and generate content that explicitly violates their safety policies. One can protect against IMPs by improving the guardrails and making them scale with the LLMs' capabilities. For two categories of attacks that are straightforward to implement, i.e., bijection and encoding, we discuss two defensive strategies, one in token and the other in embedding space. We conclude with some research questions we believe should be prioritised to enhance the defensive mechanisms of LLMs and our understanding of their safety.

♻ ★ Statistical Deficiency for Task Inclusion Estimation

Loïc Fosse, Frédéric Béchet, Benoît Favre, Géraldine Damnati, Gwénolé Lecorvé, Maxime Darrin, Philippe Formont, Pablo Piantanida

Tasks are central in machine learning, as they are the most natural objects to assess the capabilities of current models. The trend is to build general models able to address any task. Even though transfer learning and multitask learning try to leverage the underlying task space, no well-founded tools are available to study its structure. This study proposes a theoretically grounded setup to define the notion of task and to compute the {\bf inclusion} between two tasks from a statistical deficiency point of view. We propose a tractable proxy as information sufficiency to estimate the degree of inclusion between tasks, show its soundness on synthetic data, and use it to reconstruct empirically the classic NLP pipeline.

comment: 34 pages

♻ ★ Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Video Diffusion Transformer

Jiahao Cui, Hui Li, Yun Zhan, Hanlin Shang, Kaihui Cheng, Yuqi Ma, Shan Mu, Hang Zhou, Jingdong Wang, Siyu Zhu

Existing methodologies for animating portrait images face significant challenges, particularly in handling non-frontal perspectives, rendering dynamic objects around the portrait, and generating immersive, realistic backgrounds. In this paper, we introduce the first application of a pretrained transformer-based video generative model that demonstrates strong generalization capabilities and generates highly dynamic, realistic videos for portrait animation, effectively addressing these challenges. The adoption of a new video backbone model makes previous U-Net-based methods for identity maintenance, audio conditioning, and video extrapolation inapplicable. To address this limitation, we design an identity reference network consisting of a causal 3D VAE combined with a stacked series of transformer layers, ensuring consistent facial identity across video sequences. Additionally, we investigate various speech audio conditioning and motion frame mechanisms to enable the generation of continuous video driven by speech audio. Our method is validated through experiments on benchmark and newly proposed wild datasets, demonstrating substantial improvements over prior methods in generating realistic portraits characterized by diverse orientations within dynamic and immersive scenes. Further visualizations and the source code are available at: https://fudan-generative-vision.github.io/hallo3/.

♻ ★ HERO: Human-Feedback Efficient Reinforcement Learning for Online Diffusion Model Finetuning ICLR

Ayano Hiranaka, Shang-Fu Chen, Chieh-Hsin Lai, Dongjun Kim, Naoki Murata, Takashi Shibuya, Wei-Hsiang Liao, Shao-Hua Sun, Yuki Mitsufuji

Controllable generation through Stable Diffusion (SD) fine-tuning aims to improve fidelity, safety, and alignment with human guidance. Existing reinforcement learning from human feedback methods usually rely on predefined heuristic reward functions or pretrained reward models built on large-scale datasets, limiting their applicability to scenarios where collecting such data is costly or difficult. To effectively and efficiently utilize human feedback, we develop a framework, HERO, which leverages online human feedback collected on the fly during model learning. Specifically, HERO features two key mechanisms: (1) Feedback-Aligned Representation Learning, an online training method that captures human feedback and provides informative learning signals for fine-tuning, and (2) Feedback-Guided Image Generation, which involves generating images from SD's refined initialization samples, enabling faster convergence towards the evaluator's intent. We demonstrate that HERO is 4x more efficient in online feedback for body part anomaly correction compared to the best existing method. Additionally, experiments show that HERO can effectively handle tasks like reasoning, counting, personalization, and reducing NSFW content with only 0.5K online feedback. The code and project page are available at https://hero-dm.github.io/.

comment: Published in International Conference on Learning Representations (ICLR) 2025

♻ ★ Physics-Informed Diffusion Models ICLR 2025

Jan-Hendrik Bastek, WaiChing Sun, Dennis M. Kochmann

Generative models such as denoising diffusion models are quickly advancing their ability to approximate highly complex data distributions. They are also increasingly leveraged in scientific machine learning, where samples from the implied data distribution are expected to adhere to specific governing equations. We present a framework that unifies generative modeling and partial differential equation fulfillment by introducing a first-principle-based loss term that enforces generated samples to fulfill the underlying physical constraints. Our approach reduces the residual error by up to two orders of magnitude compared to previous work in a fluid flow case study and outperforms task-specific frameworks in relevant metrics for structural topology optimization. We also present numerical evidence that our extended training objective acts as a natural regularization mechanism against overfitting. Our framework is simple to implement and versatile in its applicability for imposing equality and inequality constraints as well as auxiliary optimization objectives.

comment: 26 pages, 9 figures, 3 tables; ICLR 2025 camera ready contribution

♻ ★ The Breakdown of Gaussian Universality in Classification of High-dimensional Linear Factor Mixtures ICLR 2025

Xiaoyi Mai, Zhenyu Liao

The assumption of Gaussian or Gaussian mixture data has been extensively exploited in a long series of precise performance analyses of machine learning (ML) methods, on large datasets having comparably numerous samples and features. To relax this restrictive assumption, subsequent efforts have been devoted to establish "Gaussian equivalent principles" by studying scenarios of Gaussian universality where the asymptotic performance of ML methods on non-Gaussian data remains unchanged when replaced with Gaussian data having the same mean and covariance. Beyond the realm of Gaussian universality, there are few exact results on how the data distribution affects the learning performance. In this article, we provide a precise high-dimensional characterization of empirical risk minimization, for classification under a general mixture data setting of linear factor models that extends Gaussian mixtures. The Gaussian universality is shown to break down under this setting, in the sense that the asymptotic learning performance depends on the data distribution beyond the class means and covariances. To clarify the limitations of Gaussian universality in the classification of mixture data and to understand the impact of its breakdown, we specify conditions for Gaussian universality and discuss their implications for the choice of loss function.

comment: 34 pages, 10 figures, accepted by ICLR 2025 (https://openreview.net/forum?id=UrKbn51HjA)

♻ ★ ECBench: Can Multi-modal Foundation Models Understand the Egocentric World? A Holistic Embodied Cognition Benchmark

Ronghao Dang, Yuqian Yuan, Wenqi Zhang, Yifei Xin, Boqiang Zhang, Long Li, Liuyi Wang, Qinyang Zeng, Xin Li, Lidong Bing

The enhancement of generalization in robots by large vision-language models (LVLMs) is increasingly evident. Therefore, the embodied cognitive abilities of LVLMs based on egocentric videos are of great interest. However, current datasets for embodied video question answering lack comprehensive and systematic evaluation frameworks. Critical embodied cognitive issues, such as robotic self-cognition, dynamic scene perception, and hallucination, are rarely addressed. To tackle these challenges, we propose ECBench, a high-quality benchmark designed to systematically evaluate the embodied cognitive abilities of LVLMs. ECBench features a diverse range of scene video sources, open and varied question formats, and 30 dimensions of embodied cognition. To ensure quality, balance, and high visual dependence, ECBench uses class-independent meticulous human annotation and multi-round question screening strategies. Additionally, we introduce ECEval, a comprehensive evaluation system that ensures the fairness and rationality of the indicators. Utilizing ECBench, we conduct extensive evaluations of proprietary, open-source, and task-specific LVLMs. ECBench is pivotal in advancing the embodied cognitive capabilities of LVLMs, laying a solid foundation for developing reliable core models for embodied agents. All data and code are available at https://github.com/Rh-Dang/ECBench.

♻ ★ Reinforcement Learning-Enhanced Procedural Generation for Dynamic Narrative-Driven AR Experiences

Aniruddha Srinivas Joshi

Procedural Content Generation (PCG) is widely used to create scalable and diverse environments in games. However, existing methods, such as the Wave Function Collapse (WFC) algorithm, are often limited to static scenarios and lack the adaptability required for dynamic, narrative-driven applications, particularly in augmented reality (AR) games. This paper presents a reinforcement learning-enhanced WFC framework designed for mobile AR environments. By integrating environment-specific rules and dynamic tile weight adjustments informed by reinforcement learning (RL), the proposed method generates maps that are both contextually coherent and responsive to gameplay needs. Comparative evaluations and user studies demonstrate that the framework achieves superior map quality and delivers immersive experiences, making it well-suited for narrative-driven AR games. Additionally, the method holds promise for broader applications in education, simulation training, and immersive extended reality (XR) experiences, where dynamic and adaptive environments are critical.

comment: Published in Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - GRAPP 2025 https://www.scitepress.org/PublicationsDetail.aspx?ID=LfPv9Lfiya8=&t=1

♻ ★ Predictive Prompt Analysis

Jae Yong Lee, Sungmin Kang, Shin Yoo

Large Language Models (LLMs) are machine learning models that have seen widespread adoption due to their capability of handling previously difficult tasks. LLMs, due to their training, are sensitive to how exactly a question is presented, also known as prompting. However, prompting well is challenging, as it has been difficult to uncover principles behind prompting -- generally, trial-and-error is the most common way of improving prompts, despite its significant computational cost. In this context, we argue it would be useful to perform `predictive prompt analysis', in which an automated technique would perform a quick analysis of a prompt and predict how the LLM would react to it, relative to a goal provided by the user. As a demonstration of the concept, we present Syntactic Prevalence Analyzer (SPA), a predictive prompt analysis approach based on sparse autoencoders (SAEs). SPA accurately predicted how often an LLM would generate target syntactic structures during code synthesis, with up to 0.994 Pearson correlation between the predicted and actual prevalence of the target structure. At the same time, SPA requires only 0.4\% of the time it takes to run the LLM on a benchmark. As LLMs are increasingly used during and integrated into modern software development, our proposed predictive prompt analysis concept has the potential to significantly ease the use of LLMs for both practitioners and researchers.

comment: Accepted by FSE 2025, 5 pages, 2 figures

♻ ★ ESTformer: Transformer Utilizing Spatiotemporal Dependencies for Electroencaphalogram Super-resolution

Dongdong Li, Zhongliang Zeng, Zhe Wang, Hai Yang

Towards practical applications of Electroencephalography (EEG), lightweight acquisition devices garner significant attention. However, EEG channel selection methods are commonly data-sensitive and cannot establish a unified sound paradigm for EEG acquisition devices. Through reverse conceptualisation, we formulated EEG applications in an EEG super-resolution (SR) manner, but suffered from high computation costs, extra interpolation bias, and few insights into spatiotemporal dependency modelling. To this end, we propose ESTformer, an EEG SR framework that utilises spatiotemporal dependencies based on the transformer. ESTformer applies positional encoding methods and a multihead self-attention mechanism to the space and time dimensions, which can learn spatial structural correlations and temporal functional variations. ESTformer, with the fixed mask strategy, adopts a mask token to upsample low-resolution (LR) EEG data in the case of disturbance from mathematical interpolation methods. On this basis, we designed various transformer blocks to construct a spatial interpolation module (SIM) and a temporal reconstruction module (TRM). Finally, ESTformer cascades the SIM and TRM to capture and model the spatiotemporal dependencies for EEG SR with fidelity. Extensive experimental results on two EEG datasets show the effectiveness of ESTformer against previous state-of-the-art methods, demonstrating the versatility of the Transformer for EEG SR tasks. The superiority of the SR data was verified in an EEG-based person identification and emotion recognition task, achieving a 2% to 38% improvement compared with the LR data at different sampling scales.

comment: Accepted by Knowledge-Based Systems

♻ ★ AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systems

AgiBot-World-Contributors, Qingwen Bu, Jisong Cai, Li Chen, Xiuqi Cui, Yan Ding, Siyuan Feng, Shenyuan Gao, Xindong He, Xu Huang, Shu Jiang, Yuxin Jiang, Cheng Jing, Hongyang Li, Jialu Li, Chiming Liu, Yi Liu, Yuxiang Lu, Jianlan Luo, Ping Luo, Yao Mu, Yuehan Niu, Yixuan Pan, Jiangmiao Pang, Yu Qiao, Guanghui Ren, Cheng Ruan, Jiaqi Shan, Yongjian Shen, Chengshi Shi, Mingkang Shi, Modi Shi, Chonghao Sima, Jianheng Song, Huijie Wang, Wenhao Wang, Dafeng Wei, Chengen Xie, Guo Xu, Junchi Yan, Cunbiao Yang, Lei Yang, Shukai Yang, Maoqing Yao, Jia Zeng, Chi Zhang, Qinglin Zhang, Bin Zhao, Chengyue Zhao, Jiaqi Zhao, Jianchao Zhu

We explore how scalable robot data can address real-world challenges for generalized robotic manipulation. Introducing AgiBot World, a large-scale platform comprising over 1 million trajectories across 217 tasks in five deployment scenarios, we achieve an order-of-magnitude increase in data scale compared to existing datasets. Accelerated by a standardized collection pipeline with human-in-the-loop verification, AgiBot World guarantees high-quality and diverse data distribution. It is extensible from grippers to dexterous hands and visuo-tactile sensors for fine-grained skill acquisition. Building on top of data, we introduce Genie Operator-1 (GO-1), a novel generalist policy that leverages latent action representations to maximize data utilization, demonstrating predictable performance scaling with increased data volume. Policies pre-trained on our dataset achieve an average performance improvement of 30% over those trained on Open X-Embodiment, both in in-domain and out-of-distribution scenarios. GO-1 exhibits exceptional capability in real-world dexterous and long-horizon tasks, achieving over 60% success rate on complex tasks and outperforming prior RDT approach by 32%. By open-sourcing the dataset, tools, and models, we aim to democratize access to large-scale, high-quality robot data, advancing the pursuit of scalable and general-purpose intelligence.

comment: Project website: https://agibot-world.com/. Github repo: https://github.com/OpenDriveLab/AgiBot-World. The author list is ordered alphabetically by surname, with detailed contributions provided in the appendix